Higher Transcendental FunctionsCALIFORNIA INSTITUTE OF TECHNOLOGY BATEMAN MANUSCRIPT PROJECT A. Erpfiyt, Editor W. Macnus, F. OBERHETTINGER, F. G. Tricom1, Research Associates Higher Transcendental Functions, 3 volumes. Tables of Integral Transforms, 2 volumes.ERRATA HIGHER TRANSCENDENTAL FUNCTIONS, VOL. I. P. 5, line 10; Read[1~ z/(2n+ 2)}7’ instead of [1- z/(2n + 1)" P. 12, equation (32): Read tan™'(q/p) instead of tan”! (p/q). P. 13, equation (40): Add a>0,p > 1. P, 24, line 6 up: Read Res >1 instead of Re s > 0. P. 34, line 6 up: Formula should i cil (log t)” (og m)"*! a Oe 2 (ek), 707%) instead of & P. 41, line 5: Read e P. 52, equation (4): Read 6% instead of 67", P. 52, second equation (4): Read r! instead of m!, P.61, line 6: Read from instead of form. P. 63, line 2 up: Insert | after |arg(-z). P. 64, equation (25); Read 1+a—4 instead of 1-a+b. P. 65, line 2: Insert } before the last comma. P. 70, equation (6): Read r+1—m instead of m+1~r (four times). P. 76, equation (11): On the right-hand side read ['(a +n + 1) instead of T@+n). P. 78, last line: Read F(a~a%, instead of F(a—a,. P. 86, lines 7 up and 9 up: Omit © 1)". P. 88, line 8 up: Read 2.1(24) instead of 1.5(24). P. 93, line 16 up: Read (1) instead of (13). P. 104, equation (43): Insert z between (¢~a) and F(c+ 2),2 Errata. Higher Transcendental Functions, vol. 1 P. 107, equation (36): Read I"(b) instead of ['(c). P. 108, line 15 up: Read D#D” instead of D=D*. P. 108, equations (1) and (2): Insert | before arg(1—z)|. P. 110, last line; Read (1-z)~* instead of (1-z)"'. P. 112, equation (17): Read z”? instead of zv2. P. 112, equation (29): Read z?(2—z)"* instead of z7/(2-z)"?. P, 113, equation (34): On the right-hand side read a—:b +1 instead of a—b — P. 116, line 4up: Delete 2.1(15). P. 126: Insert a horizontal rule midway between (20) and (21). P. 145, equation (23): Read [4+ ¥v—%y) instead of P(1+ 4v—Kp). P. 154, line 10: Read Re<1 instead of Re y>-1. P. 166, line 9: Read (sinv) instead of (sin v)™. P. 168, line 8 up: Read ['(v+m+ 1) instead of [@+m). P. 169, line 5: Read 0<0’<7 instead of the second 0< O<7. P. 169, last line of equation (3): Read 0< @<7/2 instead of 0<0’< 1/2, ‘9-2V on the right-hand side, andread z~* P. 179, equation (32): Insert 2 ‘ 2 instead of z*. P. 179, last line; Read w’ instead of w, and insert w before = sign. P. 182, equation (1): Read & insteadof ¥ n=0 a= z z P. 183, line 17: Read instead of —— . nl nl P. 185, equation (1): Read 2a+ 2d instead of a+ 2d, P. 189, equation (5): Insert; after ¢ on the top line of the left-hand side. P. 189, equation (7): On the right-hand side, b P. 192, line 3: Insert [ after ,F,+Errata. Higher Transcendental Functions, vol. | 3 P. 193, line 3: Read » +1; 2,1; instead of n+1,1,1;- P. 196, equation (7): Read © insteadof ©. 7 neo P. 196, equation (10): Read 1; 4k? instead of 1,47, andread 1; 1-k? instead of 1, 1—k?, P, 196, last line; Insert Jackson's theorem at the end of the line, P. 197, equation (1): Read g~" instead of q™’ P. 197, equation (12): On the right-hand side read g” instead of g". P, 197, after equation (12) add: provided the series on the right terminates and the series on the left converges. P. 199, line 3: Read 503-516 instead of 495-516. P. 206, equation (14), right-hand side: Read E(...0; x) instead of E(s1-0: 2), P. 209, equation (10), right-hand side: Read 2°?7*9 x? instead of 2°"? x, P. 212, line 7: Read min Re}, instead of max Re b,. P. 215, equation (1): Read GIF .,C-2| instead of CY? ,,(x| « P, 216, equation (4): Read (2x%) instead of (22%). P. 216, equation (9): Read Gi(x| a, b, 26—a, b+ %) instead of CRGla, 6, 26-4, 2-a+ ¥). P. 216, equation (10): Read 47" [sin (a—-b) x)’ instead of a (sin (@ — 6) a]7* 2°, P. 218, equation (28); Read [I,_,(2x")-L,_, (2x")] instead of ,_,22) 1, ax". P. 218, equation (30); Read a+6,a~-6, a instead of a+b, ab, 0 : 2 instead of GZ x a.) ae P. 218, equation (34): Read G2 (=4 Enrata. Higher Transcendental Functions, vol. I 4 4 x x P. 220, equation (59): Read —— instead of . 64 64 P, 221, equation (65): Read Zu+v, %u-v, Mp instead of y+ ky, —v+ %p, 0. P. 221, equation (69): Erase fraction rule in G7. 2) P. 222, equation (74): Read 2F, G, bs ¢3 -x)= His cn 2(- la : Tre, P. 222, equation (75), second line: Read je, -b, -c, -d xo (2{ ~b-e hot P. 258, equation (10): Insert — after first = sign. P, 262, equation (5): Read [log x + y(a)— 2y] instead of log x, P. 263, equation (14): second form of the result: Read @(c—a, c; €) ai @(1-a, 2—c; €) instead of (c— a, c;—£) and O(1-a, 2-0;-€). P. 268, equation (36): Read ® instead of ¢. P. 269, line 3up: Read ; instead of ; on the right-hand side. P. 270, equation (7), second line: Read 1—s instead of 1- P. 271, equation (16): Insert the factor ¢°~' on the right-hand side. Henk 4 the a P. 279, line 9 up: Read x instead of x’ P. 294, lines 18 to 20: Read MacRobert, T.M. instead of MacRobert, R.M. P. 295, line 4: Read Sharma, J.L.. instead of Sharma, G.L.HIGHER TRANSCENDENTAL FUNCTIONS Volume I Based, in part, on notes left by Harry Bateman Late Professor of Mathematics, Theoretical Physics, and Aeronautics at the California Institute of Technology and compiled by the Staff of the Bateman Manuscript Project Prepared at the California Institute of Technology under Contract No. N6onr-244 Task Order XIV with the Office of Nava? Research Project Designation Number: NR 043-045 NEW YORK TORONTO LONDON McGRAW-HILL BOOK COMPANY, INC. 1953HIGHER TRANSCENDENTAL FUNCTIONS, vol. I. COPYRIGHT, 1953, BY THE McGRAW-HILL BOOK COMPANY, INC. PRINTED IN THE UNITED STATES OF AMERICA Alll rights reserved except those granted to the United States Government. Otherwise, this book, or parts thereof, may not be reproduced in any form without permission of the publishers. Library of Congress Catalog Card Number: 53-5555 8910111218HD 987 Copyright Renewed, 1981, by California Institute of TechnologyThis work is dedicated to the memory of HARRY BATEMAN as a tribute to the imagination which led him to undertake a project of this magnitude, and the scholarly dedication which inspired him to carry it so far toward completion.STAFF OF THE BATEMAN MANUSCRIPT PROJECT Director Arthur Erdélyi Research Associates Wilhelm Magnus (1948-50) Fritz Oberhettinger (1948-51) Francesco G. Tricomi (1948-51) Research Assistants David Bertin (1951-52) W. Be Fulks (1949-50) A. R. Harvey (1948-49) D. L. Thomsen, jr. (1950-51) Maria A. Weber (1949-51) E. Ls Whitney (1948-49) Vari-typist Rosemarie StampfelPREFACE The late Professor Harry Bateman of the California Institute of Technology was one of those rare scientists who, responding to the interplay between mathematical analysis and physical understanding, made outstanding contributions to American applied mathematics. His contributions to aero- and fluid mechanics, to electro-magnetic theory, to thermodynamics, to geophysics, and to a host of other fields in which his adroit mathematical skills were applied, resulted in significant advances in these fields. During his last years he had embarked upon a project whose successful completion, he believed, would prove of great value to scientists in all fields. He planned an extensive compilation of “special functions” -- solutions of a wide class of mathematically and physically relevant functional equations. He intended to investigate and to tabulate properties of such functions, interrelations between such functions, their representations in various forms, their macro- and micro- scopic behavior, and to construct tables of important definite integrals involving such functions. It is true that much of this material was already in existence. However, anyone who has been faced with the task of handling and discussing and understanding in detail the solution to an applied problem which is described by a differential equation is painfully familiar with the dispro- portionately large amount of scattered research on special functions one must wade through in the hope of extracting the desired information. Professor Bateman was eminently qualified to embark on such a com- pilation, for he was unusually familiar -~ and systematically so — with existing mathematical literature on the subject; he was exceptionally adept in mathematical analysis; and he was ever conscious of the needs of the scientist who must so often use these functions. When his death cut short his work, the California Institute of Technology, in recognition of one of its great scientists, and the Office of Naval Research, in recognition of the extremely useful service such a compilation could render to both basic and applied science, pooled their efforts to continue the task initiated by Professor Bateman. ixx SPECIAL FUNCTIONS In 1948 arrangements were completed between the California Institute of Technology and the Office of Naval Research to employ at the Calif- ornia Institute of Technology four mathematical analysts of international reputation to complete Professor Bateman’s work: Professors Arthur Erdélyi of the University of Edinburgh; Wilhelm Magnus of the University of Géttingen; Fritz Oberhettinger of the University of Mainz; and Francesco Tricomi of the University of Torino. It was not long after this team began work that it became apparent that not only would Pro- fessor Bateman’s original project find its completion in their unusually competent hands, but that the activities of such a group would lead to significant mathematical investigations and advances in the general field of mathematical analysis, as well as in the more particular field of special functions. The present compendia bear undeniable witness to the success of the undertaking. The Office of Naval Research is proud of its collaboration with the California Institute of Technology, not only for erecting this lasting memorial to Professor Bateman, but also for producing what it considers a significant contribution to general science. These compendia, which have taken their roots in Professor Bateman’s “‘shoe boxes” (his repos- itory for card files) have been nurtured into mathematical maturity under the deft minds and penetrating work of the members of the international team of Erdélyi, Magnus, Oberhettinger, and Tricomi. In addition, we are pleased to have been able to render support to several young Amer- ican mathematicians who have not only contributed to these compendia but were able to avail themselves of the opportunity to work and study under the direction of distinguished scientists in a field that is sorely in need of young recruits. We feel that special thanks should be extended to both Dean E. C. Watson of the California Institute of Technology and to Professor Erddlyi; to the former, for his extremely helpful and untiring interest in seeing to the establishment and completion of this task; to the latter, for assuming, in addition to scientific participation, both the scientific administrative duties of the project and the general editorial responsibilites for the publication of these compendia. MINA REES, Director Mathematical Sciences Division Office of Naval ResearchFOREWORD ‘The late Professor Harry Bateman was one of the greatest authorities in that part of mathematics now usually described as classical analysis. His knowledge of the literature was encyclopedic and probably unsur~ passed and his ability to utilize this knowledge for specific problems was extraordinary. Research workers in difficulties would often write to him and receive, by return mail, detailed answers to their questions together with a list of references which in many cases amounted to a complete bibliography. Tt was natural for Bateman to want to make accessible in a system- atic form the tremendous amount of material which he had collected in the course of the years. His book on Partial Differential Equations (1932) was an attempt to carry out this task in a restricted field. Although the book was received with enthusiasm, and, after twenty years, is still one of the most important books on its subject, Bateman was not satisfied with this method of providing information. For a number of years he made plan after plan to organize and prepare for publication his material, a task made extremely difficult by the very breadth of the field which he intended to cover. At the time of Bateman’s death (1946) his notes amounted to a veri- table mountain of paper. His card-catalogue alone filled several dozen cardboard boxes (the famous “‘shoe-boxes”). His family, his friends, and his colleagues at the Califomia Institute of Technology very natu- rally wished to have some of this material prepared for posthumous publication, thereby erecting a monument to one of the most distin- guished and most versatile members of the faculty of the Institute. Professor A. D. Michal, for many years a friend and colleague of the deceased, undertook the sifting of Bateman’s notes. He spent several months in this herculean task, sorted out those notes which might be considered for publication and made recommendations for proceeding further with the matter. Dr.A.Erdélyi, then of the University of Edinburgh in Scotland, was invited to prepare a detailed report and proposals, and spent the academic year 1947-48 in Pasadena for this purpose. xixii SPECIAL FUNCTIONS It turned out that Bateman’s notes ranged over a wider field than even his friends had suspected and also that no single section of this wide field was in a state sufficiently advanced for immediate publication. Indeed the field was so wide that it appeared imperative to narrow it down if anything useful was to be accomplished. Notes for books on functional equations, integrals in potential theory, binomial coefficients and factorials, and many other matters had to be laid aside entirely. Of the remaining material the most important part was a projected trilogy on the higher transcendental functions, on definite integrals (especially those containing higher functions), and on numerical tables of functions occurring in applied mathematics. Since the appearance of the Index of Mathematical Tables by Fletcher, Miller, and Rosenhead, adequate information has been available on numerical tables, and so it was de- cided to concentrate on the first two parts, and these came to be called the handbook and the integral tables. The Office of Naval Research recognized the great importance of such a work by giving generous financial support to it. Thus originated what at the California Institute came to be called the Bateman Manuscript Project. The Institute was fortunate indeed, not only in being able to persuade Professor Erdelyi to remain as its Director and as Editor of the forthcoming publications, but also in securing the services of Professor Wilhelm Magnus of the University of Gattingen (now of New York Uni- versity), of Dr. Fritz Oberhettinger of the University of Mainz (now Professor at the American University, Washington, D.C.) and of Professor Francesco G. Tricomi of the University of Turin. These distinguished and internationally known scholars were assisted by a staff of younger mathematicians, The technical preparation of the vari-typescript suit- able for reproduction by a photo-offset process was in the capable hands of Miss Rosemarie Stampfel. The present volume is the first of three projected volumes on the higher transcendental functions. These three volumes will be followed by two volumes of integral tables. The California Institute of Technology wishes to express its thanks both to the family of the late Professor Bateman for the gift of his notes and of his library, and to the Office of Naval Research and especially to Dr. Mina Rees, the Director of its Mathematical Sciences Division, for the generous support they have given to this work and for the under- standing they have constantly shown for the difficulties encountered. The Institute also wishes to record its appreciation and thanks to the following persons and organizations: to Professor Michal for his pre- liminary survey of Bateman’s notes; to the University of Edinburgh forFOREWORD xiii granting leave of absence to Dr. Erdélyi; to the Rockefeller Foundation for defraying travelling expenses for Dr. and Mrs. Erdélyi on their visit in 1947-48; to the University of Turin for granting leave of absence to Professor Tricomi; to Professors T.M. Apostol of the California Institute, R. C, Archibald of Brown University, E. D. Rainville of the University of Michigan, Mr. S. O. Rice of the Bell Telephone Laboratories, and Professor C. A. Truesdell of Indiana University for information or con- sultations in connection with the work; and to the McGraw-Hill Company for technical advice and publication. Last but not least, acknowledgments should be expressed to Dr. Erdélyi and the staff of the Bateman Manu- script Project for the faithful and highly competent performance of a difficult task. E. C. WATSON Dean of the Faculty California Institute of TechnologyINTRODUCTION The work of which this book is the first volume might be described as an up-to-date version of Part II. The Transcendental Functions of Whittaker and Watson’s celebrated ‘Modem Analysis’. Bateman (who was a pupil of E. T. Whittaker) planned his “Guide to the Functions” on a gigantic scale. In addition to a detailed account of the properties of the most important functions, the work was to include the historic origin and definition of, the basic formulas relating to, and a bibli- ography for all special functions ever invented or investigated. These functions were to be catalogued and classified under twelve different headings according to their definition by power series, generating func- tions, infinite products, repeated differentiations, indefinite integrals, definite integrals, differential equations, difference equations, functional equations, trigonometric series, series of orthogonal functions, or integral equations. Tables of definite integrals representing each function and numerical tables of a few new functions were to form part of the “Guide”. An extensive table of definite integrals and a Guide to numerical tables of special functions were planned as companion works. The great importance of such awork hardly needs emphasis. Bateman’s unparalleled knowledge of the mathematical literature, past and present, and his equally exceptional diligence, would have made the book an authoritative account of its vast subject, and in many respects a defin- itive account; a Greater Oxford Dictionary of special functions. A realistic appraisal of our abilities and of the time at our disposal led to a drastic revision of Bateman’s plans. Only Bateman himself had the erudition to give a reliable and accurate history of specia] functions, and the manpower available to us was insufficient for the inclusion of all functions. Thus we restricted ourselves to an account (probably far less detailed than that planned by Bateman) of the principal properties of those special functions which we considered the most important ones. The loss thus caused to mathematical scholarship is great, regrettable, and final but we venture to hope that it will be counterbalanced in some measure by the considerable reduction in size of the book, and by the gain in the clarity of its organization. We can only hope that although xvxviii SPECIAL FUNCTIONS NOTATION, REFERENCES The notation presents peculiar difficulties. There are special func= tions, for instance Bessel functions of the first kind, for which there is a generally accepted standard notation. There are others, like con- fluent hypergeometric functions, for which there are several essentially different and independent notations. The most awkward problems present themselves with those functions for which more or less the same symbol is used with several different meanings. Hermite polynomials are usually denoted by H, (x) or He,(x), but this symbol sometimes refers to the polynomials derived by repeated differentiation of exp(-x?), and some- times to those derived from exp (~%x7). Moreover, some authors include, others exclude a factor, n!, We attempted to use the same notations throughout our books The most significant deviation from this principle is in the case of the confluent hypergeometric series for which the symbol ,F, is used mostly, except in Chapter VI (and some of the later chapters) where the symbol © is used for the same series (and ¥ for a second solution of the confluent hypergeometric equation). Wherever possible we followed standard notations. In the case of Bessel functions we adopted the notations used by G. N. Watson in his monumental work, in the case of orthogonal polynomials we used Szegé’s notation (with the exception of using CY for ultraspherical polynomials). With Legendre functions, we followed Jaknke-Emde, Magnus-Oberhettinger and some other authors in making a distinction between the definition of Legendre functions appropriate for the interval (-1, 1) and the defi- nition appropriate for the complex plane outside of this interval. In cases of doubt we usually decided upon that notation for which more convenient or more extensive numerical tables were available. We adhered to definitions used in numerical tables even in cases in which we thought that a different definition would be preferable from the mathematical point of view. All notations are explained where they occur for the first time. There is at the end of this volume an Index of notations which will help the reader to find the meaning of any notation used in the book, and a Subject index which gives the notation for any function which occurs in the texts Many of our chapters may be read independently of the others, yet there are many cross references. Equations within the same section are referred to simply by number, equations in other sections areindi- cated by the section number followed by the number of the equation. Thus (3) means equation (3) in the section in which the reference occurs, 2.1(3) means equation (3) in section 2,1, References to the literatureINTRODUCTION xix have the name of the author followed by the year of publication. They invariably refer to the list of references at the end of the chapters The size and complexity of our compilation make it vain to hope that errors of judgment, or mistakes have been avoided. The undersigned will be glad to receive corrections or suggestions for the improvement of the work should a second edition become desirable. In conclusion I should like to express the thanks of the entire project staff to the California Institute of Technology, and especially to Dean E. C. Watson, for initiating this work and for the great understanding they have shown for the numerous problems we encountered. I should also like to thank my colleagues without whose assistance the present work could not have been carried out. A. ERDELYICONTENTS PREFACE 0 FOREWORD ...... Se cmCeecmE ew iaerMU ne veces ae INTRODUCTION... . . Ge ee ooo Go ego. CHAPTER I ‘THE GAMMA FUNCTION il Definition of the gamma function... 2... 7 12. Functional equations satisfied by '(z). . 2... 1.3. Expressions for some infinite products in terms of the gamma function... 1 1... ee eee ee 1.4. Some infinite sums connected with the gamma SUCHIN 15. he beta function, 2) 1.5.1. Definitie integrals expressible in terms of the beta FUMCHON ee oe 1.6. The gamma and beta functions expressed as contour integrals. 6 6 eee ee ee ee The function... 2... ~ eee scare aete 1.7.1. Funetional equations for W(z). . 2... ee : 1.7.2. Integral representations for ¥(z) .... 2.2 ee 1.7.3. The theorem of Gauss... 1. eee ee eee 1.7.4. Some infinite series connected with the Y-function . . 1.8. The function G(z). 2... eee ee ee ee eee 19. Expressions for the function log (z).. 2.2... 1.9.1. Kummer’s series forlogM (2)... .... oe 1.10, The generalized zeta function. » . 2... 1 ee ee Ll, The function P(z,s,0)= E (en)e2" 2... ee LILI. Euler’s dilogarithm 2 eee ee The zeta function of Riemann. . . . . Pa aa 1.13. Bernoulli’s numbers and polynomials .. 2... 2. 1.13.1. The Bernoulli polynomials of higher order... . . La. Euler numbers and polynomials... 2... 2... e xxi 13 15 16 16 18 19 20 23 24 27 31 32, 35 39 40xxii 1.14.1. 1,15. 1.16. 1.17. 1.18. 119. 1.20. 1.21. 2.1, 2.1.1. 2.1.2, 2.1.3. 2.1.4, 2.1.5. 2.1.6. 2.2. 2.2.1, 2.2.2. 2.3. 2.3.1, 2.3.2. 2.4. 25. 2.5.1. 2.5.2. 2.5.3. 2.5.4. 2.5.5, SPECIAL FUNCTIONS The Euler polynomials of higher order... 2... Some integral formulas connected with the Bernoulli and Euler polynomials... . . . aaa Polygamma fincticis 502 Some expansions for log (1+z), w(1+z), G(1+2), end a) ee ere ea seeds ae Asymptotic expansions. . 2... ee eee ‘ Mellin:Vames integrals Power series of some trigonometric functions. . . . . Some other notations and symbols... 2... +2. References cs ee eee fe CHAPTER IL THE HYPERGEOMETRIC FUNCTION FIRST PART: THEORY The hypergeometric series... 2. 1. ee The hypergeometric equation. . 2... 1. ee Ae Elementary relations. 2... ee eee eee The fundamental integral representations... . . . Analytic continuation of the hypergeometric series. . . Quadratic and cubic transformations. . 2... 1. F(a, 6; c; z) as function of the parameters... . . . The degenerate case of the hypergeometric Cen A particular solution... .- 6. +--+ - The full solution in the degenerate case... 1... The full solution and asymptotic expansion in the peuctal case oe Linearly independent solutions of the a equation in the non-degenerate case... . . ee Asymptotic expansions, . . . . Integrals representing or involving hypergeometric functions) 2 ee ee Miscellaneous results . . . A generating function... 2... oe Products of hypergeometric series. . 2... 26 ee Relations involving binomial coefficients and the incomplete bera function A continued fraction. . 2. ee ee ee ee ee ee Special cases of the hypergeometric function... . -2.6. 2.6.1. 2.6.2. 2.7, ae 2.7.2, 2.7.3. 2.7.4. 2.8. 2.9. 2.10, 2.12, 3.6.1. 3.9.1. 3.9.2. 3.10. 3.1L. 3.12. 3.13. 3.14. CONTENTS Ricmens’s equation Reduction to the hypergeometric equation»... ~. + + Quadratic and cubic transformations. .. ...- +--+ Conformal representations... . 1.1 ee eee Group of the hypergeometric equation... . oe Schwarz’s function. 6. 6 6 ee eee ee ee ee Uniformization. » . - ee eee ee ee CeO eee eee ee SECOND PART: FORMULAS The hypergeometric series... ee ee ee ee Kummer’s series and the relations between them - Analytic continuation. . . Quadratic and higher transformations... . 2. 6s integrals) eae as References... 0.2.0 - eee eee reece CHAPTER III LEGENDRE FUNCTIONS dntrodectiony (20. @ 0 cece ee ee eee The solutions of Legendte’s 's differential equation, . . . Relations between Legendre functions... . 1.6 6 Some further relations with hypergeometric series. . . Teccoke functions onthe cee : Trigonometric expansions for P#(cos 6) and Ce Special values of pandv .. 1... ee ee ee eee Teseuke polynomials integral represeuteons Relations between contiguous lemiae acne | Asymptotic expansions . 2. 1 6. 1. ee eee Behavior of the Legendre functions near the rn a Eine tends eee The addition theorems. . - - - - See ccs Integrals involving Legendre functions ... ~~... - ‘The ring or toroidal functions... .-. 5... oe The conical functions. «676 + ee ee eh ee ee ee 101 105, 108 110 114 il? 120 140 141 143 146 148 150 155 160 162 163 165 168 169 173 174xxiv SPECIAL FUNCTIONS 3.15. Gegenbauer functions... ......2020- 175 3.15.1. Gegenbauer polynomials . . 2.2... 2. ee ee 175 Ab Gepeubener functions © 178 3.16. Some other notations... 2... ee ee eee 179 References. . » s 0 + ee ee eet ee tee 180 CHAPTER IV THE GENERALIZED HYPERGEOMETRIC SERIES 41. dneodMctiONs © ees cc 182 4.2. Differential equations -.. 21... ee ee ee 184 4.3. Identities and recurrence relations. . ... - + + . 185 44. Generalized hypergeometric series with unit argument in the cesep=gt+l.....- Bean 188 4.5. Transformations of |,,F, and values for arguments other than unity... 1... ee eee 190 4.6. dtepele 192 47, | Vorousspecialresilin | 192 4.8. Basic hypergeometric series . . 2... Bae 195 References... -. +++. Poe ee eee 199 CHAPTER V FURTHER GENERALIZATIONS OF THE HYPERGEOMETRIC FUNCTION ele Various generalizations ..........2-% 202 MACROBERT’S E- FUNCTION 5.2. Definition of the Z-function -..... + - wae 203 5.2.1. Recurrence relations. see 205 5.2.2. Integrals... ... : MELJER’S G- FUNCTION 5.3. Definition of the G-function ........ 206 5.8.1, Simple identities ©... 66-662 ee eee 209 5.4. Differential equations»... . 1... ee ee 210 54.1, Asymptotic expansions... ......-..+- 211 So Seiesendigeeale eis Se eee 213, ee eee cree 214 5.5.1. Series of G-functions . . 5.5.2. Integrals with G-functions .CONTENTS bad 5.6. Particular cases of the G-function .... . or 215 HYPERGEOMETRIC FUNCTIONS OF SEVERAL VARIABLES 5.2, Hypergeometric series in two variables . 2... 222 Oe 224 5.7.2, Convergence of the series... ....---4- 227 5.8. Integral representations .. 2... 2 eee ee 229 5.8.1. Double integrals of Euler’stype ......... 230 5.8.2. Single integrals of Euler’s type... 2... e281 5.8.3. Mellin-Barnes type double integrals... ..... © 282 5.9, Systems of partial differential equations ..... . 232, 5.9.1. Ince’s investigation. . 6... eee ee ee es 237 5.10. Reduction formulas... 2. ++. eee eee 237 5.1L. Transformations. . . se 239 5.12. Symbolic forms and expansions 7 243 SiS Specislccee, a 5.14. Further series»... 2-2. oe References... ....-.. oo CHAPTER VI THE CONFLUENT HYPERGEOMETRIC FUNCTION 6.1, Orientation. 6 6 6 1 ee eee ee ee 248 6.2. Differential equations . . . fe 249 6.3, The general solution of the confluent equation near the origin 2... ee ee ee ee ee 52 6.4, Elementary relations for the ® function ...... 253 6.5. Basic integral representations, ......... 255 6.6. Elementary relations for the ¥ function ...... 257 6:1: Fundamental systems of solutions of the confluent Cte 258 Ube logarithmic case cg Further properties of the ¥ function eee eae 262 Whittaker functions, .. 2 26 6 1 te ee ee 264 Bessel functions, .. . 1... ee ee ee es 265, Other particular confluent hypergeometric functions. 6 6. ee ee ee ee see 266 6.10, Laplace transforms and eonfigent hypergeametric functions © 6 eee ee ee ee ee 269 Gil teraleerescuitions © O71 Gill: The@ Gainxxvi SPECIAL FUNCTIONS 6.11.2, The ¥ function... 2... +2 e ee ee eee 6.11.3. Whittaker functions .. 2... Beetles carars 6.12. Expansions in terms of Laguerre polynomials and Bessel functions... . . ee 6.13. Asymptotic behavior. .. 2... - a 6.13.1. Behavior for large zx]... ee ee ee 6.13.2. Large parameters... eee 6.13.3. Variable and parameters large... 2 2 ee ee 6.14. Multiplication theorems. . . . 6.15. Series and integral formulas G51 senes 6.15.2. Integrals... . Be ee 6.15.3. Products of — hypergeometric functions . . 6.16. Real zeros forreala,c.... - eee 6.17. Descriptive properties forreala,c,x .. 1.5 References... 6 2 ee ce ee ee ee eee SUBJECL INDEX 0 0 INDEX OF NOTATIONS © +e ee eee eee eee ee ee 273 274 275 277 278 278 280 282 283 283 284 286 288 291 293 296 301CHAPTER I THE GAMMA FUNCTION 1.1. Definition of the gamma function The function '(z) can be defined by one of the following expressions: (1) They= fe tet de =f) (log 1/e)*7 de Re z>0, nin? 2 (2) TG) = lim a. Fase re” , Sasnaeea (+ Vn)? Q+2/)74, (3) YPG)=ze%* i (1 4z/n)e "I, n=1 where (4) y= lim ( ¥ 1/m- log m) = 05772156649 +++ noo nat denotes Euler’s or Mascheroni’s constant. The definition (1) was used by Euler, (2)(in a slightly different notation) by Gauss, and(3)by Weierstrass, Replacing ¢ by st in (1) (s real and positive) we get (5) Pajas? {Qe tee! de Rez>0, It can be shown [cf. 1.5(34)] that this formula holds for complex values of s and for a path of integration along the straight line from the origin to we ©, Thus we have © Gast J" enti de -(4n+ d)0. This equation holds for arg s + 8 = +}4n provided 0< Re z <1. a2 SPECIAL FUNCTIONS From (2) and (3) it is seen that the gamma function is an analytic func- tion of z whose only finite singularities are z = 0, -1, -2, -... From (1) it follows that (7) PG) = fi evter des fP emt et" de =P (2) + Ola), Q(z) being an integral function, Expanding e~‘ in a power series and inte - grating term by term: (8) P(z)= ES Cy ial (esa). F = Hence it follows that (-1)"/n! is the residue of I'(z) at the simple pole z=-n,(n=0, 1,2...) [cf L171]. It will be shown that the expressions (1), (2), (3) represent the same function. For a positive integer n and Re z > 0 repeated integration by parts yields nln? SO C2 Wes Det esa) so that by Tannery's theorem - Hep ace ye tela Ii : eee Thus (1) is equivalent to (2). Equation (3) can be deduced from (2) as follows. By (2) we have VT (z) = lim 2(1 + 2)(1+4 Ye) oe (14 2/n) e778" n= 00 or VV (z) = lim[z (14 2) e7? (+ 4z) eo #5 (14 c/n) ee 2 nso we ttt KE eee tn lag nly and finally WE(z) = 26% Th C(1+ 2/n) eo 1, nA If the real part of z is negative, andn + 1>Rel-z)>n,(n=0, 1, 2,...), T'(z) can be represented by an integral due to Cauchy and Saalschitz (Whittaker-Watson, 1927, p. 243): (9) Pi)= Pet & eye/mi ert at ~(v+ )0, Rey>0. Substituting ¢ =v/(1+v), the relation (2) Bz y) = a (40) 7 dv Rex>0, Rey>0 ¢ he fates is obtained, and from this [erst (8) Boy 9) = {1 © +04 yA dv Ho eo can be deduced. It follows that (4) Bis, y) = BG, x). ff Te+y) “ote ane a taro [cf. 1.1(4)] is multiplied by v*~', integrated with respect to v between 0 and «, and if the order of integration is inverted, we have SP ae Oe eae doe Gey) vee (Le ae de or PWPG) Te+y) ’ the expression for the beta function in terms of the gamma function, The following functional equations for the beta function can be deduced easily from (4) and (5). (See Section 1.2): (6) Bay + D=(/2) Bee + 1, = ly/(x + IB y), (7) By y) Be +y, 2) = Bly z) BY + 2,4) = Bz, x) B(x +2, y), TPG) Pw) Ti+y+z+u) 1 ntm=1 1) (9) mea” ( | )- a( oF 1, m, positive integers. 1.5.1. Definite integrals expressible in terms of the beta function By means of suitable substitutions, a number of definite integrals, such as the following, are reducible to the beta function: (5) Be, y) = (8) Bi, y) Be +y, 2) Ba +y +z, u)=10 SPECIAL FUNCTIONS 1.5.1 (10) Bay) = 2° {Lads 4 or] de Rex>0, Rey>0, ay Us ee} (L-2)77! (1+ bt) 477 de = (14 6)7* Bx, y) b>-1, Rex>0, Rey>0, (12) fer! (14 be) #77 de = b™* Bs y) b>0, Rex>0, Rey >0, (13) f° - 8)="1 (a= 277! de = (a - 6)**77! Bw, y) Rex>0, Rey>0, b0, Rey>0, c0, Rey>0, b0, 6>0, O0, Rey>0, Rex>0, an J, (a) f) (Le ert L~ o8 (4 28) ae = 2449"? BG) Rex>0, Rey>0 Substituting trigonometric and hyperbolic functions, we obtain a number of integrals involving trigonometric and hyperbolic functions: (19) S.%" (sin 7" (cos oP" de = % Bl, 9) Rex>0, Rey>0, ; Rex>0, Rey>0, 5>-1,15.1 GAMMA FUNCTION n Mor (oe tent yt (21) f ee) (cos? t +b sin? 2F*7 Rex>0, Rey>0, b>0, (22) pe cosh ¢ (sinh #7" (145 sinh? t)"*"¥ de = %57* B(x, y) Rex>0, Rey>0, 6>0, (23) JP (sinh 2 (cosh 2)"8 dt = ¥B(ha + %, YB ~ Ya) Rea>-1, Re(a-f) <0, (24) SP eH (L—e #7! de = 27! Blx/z,y) Rex/z>0, Rez>0, Rey>0, (25) f° e*E sinh (Be)]7 de = B' 2-7 BOF F~ Fy, Lt y) Rey>-1, Ref>0, Re(a/B)>Rey, * cosh (2at) py 2 = a/p, B- (26) [a Garett 41 PBB + a/p, B~e/p) Re(8 ta/p)>0, p>0, (22) JP eos (22t) sech (nt) de = % sech z [Im 2] < Ka, (28) J cosh (221) sech(nt) de = sec z [Rez| <%a, Formula (27) is known as Ramanujan’s formula. Formulas (12), (13), (17), (19) originate from (1); (11) from (2); (10) and (26) from (3); (14), (15), (20), and (21) from (11); (16) and (22) from (12); (18) from (16); (24) from (17); (23) from (22); (25) from (24); (27) and (28) from (26); all are obtained by easily recognizable substitutions or specializations of parameters. Evidently the range of validity of the formulas (11), (20), and (12), (16), (21), (22) with respect to 5 can be extended to any values of 5 in the complex b-plane supposed cut along the real axis from~—1 to ~ 2 and from 0 to ~ © respectively. By complex integration it is possible to express some further trig- onometric integrals in terms of the gamma function, Consider L(t 28s where C is a contour consisting of the upper semi-circle |z| = 1 and its diameter, The contour is indented at z = 0, + 1, and the radius of each12 SPECIAL FUNCTIONS 15.1 indentation is ¢. On letting « approach zero, one obtains (cf, Nielsen, 1906, p. 158) the following result : 7 TU +a) ae aaa ace eee cee ee 2 TOs S42) ras) * (29) f "(sin t)%e #* de= ° Rea>-1. If C is a contour consisting of the semi-circle |z| = 1 in the right-half plane and the straight line joining the points z = +2, with indentations at z= 0, +i, and if the radii of indentation are made to approach 0, the evaluation of Slat + 3t2P" de gives 7 T+) % 30) 2% t) di (30) J,*” (cos 4)*cos (Bd) de Ras ras Rea>-1l, For other similar integrals see 24(6) to 24(10). Next consider fe z@*e Sdz c>0, where the contour C consists of the real axis from + ¢ to +R, the arc of the circle z = Re‘? from $= 0 to $=A(-%7 < B < 4m), the straight line from z = Re¥ to ee ®, and the arc of the circle z = ce*? from B= Bto @= 0. Since the value of the contour integral is zero, on making € > 0 and R > & it follows that (31) if pant pret cosf-ietsin B gy _ P(g) o~ tet a8 ~'4nO, or B=t%n, 00, Rea>O, or p=0,' 00, Res>0O or Res=0, 00, ~(47+8)0, Rea>0, -a0, Rea>~l, -Mr0, p>l, (40) f° sin (at?) dt = (pa%®)“* P(1/p) sin (a (2p)7"} 1.6. The gamma and beta functions expressed as contour integrals We use the notation Serre dt for an integral taken along a contour C which starts at a point ¢, encircles the origin once counter-clockwise and returns to its starting point, it being understood that all singularities of the integrand except ¢ = 0 are outside C. Consider [{°*) ete dt, the initial and final values of arg t being ~ 7 and +7 respectively. Taking C to consist of the lower edge of the cut from ~% to~p, the circle ¢ = pe"?(-7 < @ < m), and the upper edge of the cut from ~p to ~~, we find that SRP | bent de =2i sin(n2) [Perv # dv +L, where I denotes the integral along the circle (t{ = p. Since I tends to zero with p, provided that Re z < 1, we have, in view of 1.1), Q) [2% et 2 # de = 28 sin (zz) P(L— 2) or, by means of 1,26), Hankel’s representation (2) UVa) = 12nd) [EP et oF de larg t] 0, fOtOF I VO-d pet (y_ yet dew (127%) (1- e279) B(x, y) Rex>0, Rey>0, Hence (4 oF, 1-,0-) _evinaty) (7) Bly y)= el n ds, 4sin(rx) sin(ry)7 GAMMA FUNCTION BL and by the theory of analytic continuation this formula, derived in the first instance for Re x > 0, Re y > 0, x y, not integers, holds for all values of x and y except integers. It is due to Pochhammer. In a similar manner B(x, y) can be represented as a single loop integral (8) Bla, 9) =% esch(wiy) [UP 271 (2 - DT" ae Rex>0, |argt-Di| <7, yA0, +1, 2,..., (9) Bx, y) =-% esch(nia) JO" (71 (1-2)! de Rey>0, |arg(-O! <7, «#0, 41, 42,-.-- 1.7. The % function The function y(z) is the logarithmic derivative of the gamma function: dlog@(z)_ I''(z) dG) From equations 1.1 (2) and 1.1(3) we obtain the representations 1 1 1 a eo in| Q) ul) = or log I*(z) = J, *y(z) dz Q) Y= lim [ me @) wl) =-y~Q/a) +B s/n em =-y+-D © Vint D+me ano The w function is meromorphic with simple poles at z = 0, -1, -2,.... Clearly @ wQ=-y. From equation 1,3 (2) with u =z, v = 1 we have ra+2 2 : (5) log = log z =-y + F Un 4 71 = fog + Vint adi. Te - From equations (3) and (5) (6) ba) =logz- & tn4 7! ~logll+ Vint DL oe From equation (6) and 1.1 (1) () yaw) =F [oo = tog dea) = Je" og t dt16 SPECIAL FUNCTIONS L721 1.7.1. Functional equations for ¢(z) From equations 1.2(1), 1.2(2), 1.2(6) and 1.2(11) we have (8) Yl) =p 42)- Vz, (0) vGenele tet UAsm=le t+ 5+ 1 tee z+1 z+ Go uc 2 Zt U) meL Ben, - (QD de) - pl -z) =~7 etn(az), wz) ~ W(-z) =m etn (az) - Vz, wl + 2) — Wl - 2) = 27-7 etn (nz), 04 +z) — (4 - 2) = 7 tan(z), (12) (mz) = m7!" Wz + r/m) + log m. i 1.7.2, Integral representations for 1/(z) The formula (13) wa) =ny+ f= NG a tae Re z>0, is easily verified by expanding (1 ~ #)~' into a series, integrating term by term, and using (3). The substitution ¢ = e~* gives (14) Yl) eye ete De ade Rez>0. Hence we have wl $+ $ (a+ B)/b]- lz + $@-Bf)/b) = 2b [ e°** sink 80 [ sinh 617" dt Relat +f)>0, From (11) we obtain a formula for (2) valid for Re # <1, (13) (2) =—y -7 etn(ne) +f (1-874) = 0) de Rez<1 7 (16) Wz) == y —a etn(rz) + Le (et D7 de Rez0,1.7.2 GAMMA FUNCTION ccd can be proved as follows. Integrating 1 o-xt x o e*t dt with respect to x from 1 ton we have (18) logn = f° (et= e7"*) 7" de. Introducing this and I/(z + m) = [© e~"*** de in (2) we have W(z) = Tim CLE (e7 te) 1 em HEE wee omelet HN] ap} = lim {fe ete eT de - deel (L-e7 978 8 E4*Y] deh The first integral is independent of n, and the second tends to zero as n » 2, This proves (17). Taking z = 1 in (17) an integral formula for Euler’s constant is ob- tained: (19) y= dees) -t et de, With ¢ = log(1 + 2) and 8 = log(1 + A) we have from (17) W() = Tim Sle ee 8c] de : = Jin f . fe ee SF Jin, SS de + lim, fPLe (4 2) at dee Since the first limit is zero, Dirichlet’s formula, (0) Plz) = [Let (L4 dF] 8 de Re z>0, follows. Also we have (2) y=- WD == JP fete at =- JSP Leost- (140?) 7" de, ° The first integral follows from equation (20), andthe second can be obtained by integrating ~' e~t — ¢7'(1+ 2)~' around a quadrant of a circle in- dented at the origin, the origin being the centre of the circle. From equations (20) and (21) we obtain Wha) =-y + SU 4 =e et de Rez>0,at SPECIAL FUNCTIONS 17.2 Binet’s expressions, (22) (2) = log 24+ [Pe '~ (=e) ede Rez>0, (28) Ula) = loge - Ke ~ (PL e 9 -et-K eo ade Rez>0, (24) We) = log z + f° (= et)! 4 7 = en de Re z>0, (25) U(z) = log z— 427! ~ {et '- 0's lew de Rez>0, can easily be obtained from (17) and (18). ‘The more general expression (25) Wl) = log =e" Let Wa ed eat hn < B 0, (27) Wz) = which is likewise due to Binet. Hence we have (28) y=-Y(D = 44 QS 2+ (eM Dede, 1.7.3, The theorem of Gauss Taking z = p/q in (13), 0

‘cos(2pn/q) logl2 - 2 cos(27n/q)]. The prime indicates that in case of an even g only one-half of the last term shall be taken in the sum. Thus for a positive proper fraction z the value of y(z) can be expressed as a finite combination of elementary functions. By means of (10) this result may be extended to every rational value of z. This is Gauss’ theorem, 1.7.4, Some infinite series connected with the y function If we define Af) =fle+D=f, A*f(2)= § (A)ieena ens it follows from 1,7(8) that Avla+z)=V(a+2), so that we have A? datz) = AL1/(a +2)] =~ 1/[(a +2) (a+24+)] and A™y(a +2) =A™' [V/(a+z)] =D" (n-D!/[(atz) (a+z4+ Dos (o+z4+n-D Hence the development of ys(a +2) in a factorial series is convergent for Re(a + 2) > 0, @ not a negative integer, and is of the form (Nérlund 1924, p- 261) 2 1 az-2) 1 ez-D(z-2) us =¢ ——— frog ied Eeuec nL OO ee seen s a ee) The functional equation 1.7(10) is useful for summing some series. We have for instance: (31) 2 (a +mb)7' =b7! 3 (m +a/b)7' = 67" [hb (n+ 14+0/b) -y (a/d)],. 2 = 1 1 ue :$ ‘a—b “Wa oe eh ae oh com eas 2 ce) lon a 1 a a a wuvsfo(gpened)-o(Se3-o(Gened) Ge}2 SPECIAL FUNCTIONS 17.4 and, ifn +e, 1 1 1 SS a+b a42b a+3b mee = BOWL ab mr (444 aby, 18. The function G(z) ‘The function G(z) is defined by (CD) G(z) = Uh + ez) - WChz) From 1.7(13)and L7(14) we have (Q G@)=2f 4d de Rez>0, (3) G)=2 fe (sey de Rez>0, A consideration of { e~##(1+ 7 *)~' de extended over the contour used in deriving 1.5(31) yields the more general representation Clare ease arta ~ Yin 0. Since (Whittaker-Watson, 1927, p. 249) (3) le mth + (et 17]! eo tde = 1—% log(2z), we have Binet’s first expression of log I'(z), (4) log P(z) =(z —%) log z —z +% log(27) + fPllet- Yt a yet e Bade Rez>0, or, more generally [ cf, 1.5(1) and also 1,7(25), 1.7(26)], (5) logI"(z) =(z — 8) log z~z +% log(2n) + Jee ett art ay et en Bde ~kn0. From 1.7(8) we have a log Mg dz? (10) & (2) = = Viz+2)% azo Now we make use of a summation formula due to Plana (Lindeléf, 1906, p61), Cay Ef) = (C0) + J flr) devi LC pled ~ fa" DT a, valid if Ll) f(2) is regular for Re £ > 0, C=rtit, 2 lim eo ldl f(r 448) = 0 uniformly for 0. 0) in (11) we find that (2) a Va tna oz) = eta! + LP hee? 4 2°08 (MAIS ae. Integrating twice from 1 to z we obtain (13) log P(2) = (z - 4) log z+ 2(A- 1) +B + BSP (P— N tan™ (¢/2) de, A and B being integration constants. To determine these, we note that 0 < tan"'x 0 so that | log P(z) - (@ ~¥) log ¢ -(4- Dz ~8/ < (a) J (em Died for z real and positive. The right-hand side vanishes as z + » through positive real values, and by comparison with (8) we at once have A = 0, B ="%log(2z). This proves (9).19.1 GAMMA FUNCTION 23 19.1 Kummer's series for log I'(z) ‘The function log!’(a), 0 < x < 1, can be expanded in a Fourier series. We chall use the known Fourier expansions (Bromwich, 1947, pp. 356, 393, and 370 respectively): log(sinns) =~ log 2- ¥ (/n) cos (2anx), esch(4e) sinh(4~2)t~= 8a Z [n sin(Qma)]/(e? + 42? 0%), (1-22) = 2 5 (/n) sin (2mx). If these are substituted in (7) with z = x, we have to evaluate the integral f c= L1G cero) He 22447? n? 2m 1+ 2? t a . aL GG promt) [ siiitinia [lone of] “Qan lo t and by means of 1,7(21) and 1,7(18) this is (2m)7'[y + log(2ma)l since we have for the third integral : jim, Sp (cost e"") 7" de = Jim, (Ei(-6) - Ci(6)1=0. Thus we have (14) log P(x) = % log(27) + £ [(2n)"! cos(2mmx) +(y + log 2m) (m)7' sin(2ma)], mat log P(x) = (4 ~ 2) (y + log 2) + (1-2) log # ~ flog (sin na) (m)7' logn sin (2mnx) O by the equation DM Cls9= 2 wane v#0,-1,-2 0. It satisfies the functional equation D yls,d ays, med +S) say oe Since for Re s >0 and Re v > 0 we have from 1,1(5) (a+ a)8 PGs) = em de,Ly GAMMA FUNCTION 8 it follows that (3) T(s) dls, 0) =i peer ect (le eat)ay de =f) 2°" (1-2) og Vx)! de Res>1l, Rev>0. Considering f ¢*' e7*(1~- e797! de taken around the complete bound- ary of a sector of a circle, indented at the origin [ cf. L5(1)], we have the more general representation woeth A) TAs) elayw) = fe eM eI at Res>1, -MnO, larg) < a This integral gives a representation of ¢(s, v) valid over the whole s-plane with the exception of the points s = 1, 2, 3,.... From it Hurwitz’ series representation of ¢(s, v) can be obtained. Consider Soret e i tae taken around a closed contour C starting at the point ¢ = (2V + lm and tem consisting of a circle K and a loop L as indicated in the above figure.6 SPECIAL FUNCTIONS LD ‘The radius of the circle is (2N + Im (V an integer), and the loop L does not contain any of the points ¢ = + Qni, + 4m, + 6ri, .... In the region bounded by C the integrand of (5) is analytic and one valued except at the simple poles + 2ni, t47i, ... , +2Nmi, By the theorem of residues fcget ert f[ ae | ee = K L where R, and R,! are the residues of the integrand respectively at 2nzi 1 gout L and ~ 2nzi, Ry = (ana)t1 eB gmAM IVE RY a (Ian)! @ CHT (SN) genes, Letting N + « we find that the integral over K tends to zero provided Res < 0 andO < uv < 1. By means of (5) we thus obtain Hurwitz’ formula © Clsa) = 2m ~s) ¥ nt sin (mv + 4s) Res<0, O 0, which is Hermite’s representation of ¢(s, v). From (7) it can be seen that ¢(s,v) has only one singularity (a simple pole with residue 1) in the finite part of the s-plane. Furthermore we have Cef 1.7(27)) (8) ¢(0,v) =% -v, : i a peaeeaeee dt @ sin [9-2 ] : ge eer? [ae = ia) Rev> 0. Differentiating (7) with respect to s, then putting s = 0, and using 1,9(9) we obtain da (19) (2) = log Mv) ~ % log(27). ds |saoLl GAMMA FUNCTION 2 In the special case when s = —m(m = 0, 1, 2,.+.), we have By, @) m+l where B, (v) denotes the Bemoulli polynomial { cf, 1,13(3)]. To prove this, we note that if s is an integer, the integrand of (5) is a one-valued function of t, and we may apply Cauchy’s theorem. If s = ~m(m = 0, L, 2, v++), we have [ cf. 1.13(2)] mot mt en ee (2) ee et Gare! ape GD oe prest =(-ps"" ie Cys, @) ——. n! 0) Thus the residue of the integrand at ¢ = 0 Pee , and this proves (LD. L141. The function ®(z, s,0)= E (v+n)"*2* amo The function (DO, 5,0) = EF (went 2* ll <1, v#0,-1L-2 n=o satisfies the equation (2 ylasye)= 2" ylasm+0) +S! (ve ade 2 m= 12,3606, v#0,-1-2,.... Since (+ ny7t eM A LY/P(S)] LP ett et (ze 9" de Rev>0, Res>0, Lope tener 3) (2,5, | —— ————- ae ca aoa Hee 0 dete) Lz# LRes>0 orz=lRes>l From 1.1(5), we have the integral formula If a cut is made from 1 to © along the positive real z-axis, ® is an analytic function of z in the cut z-plane provided that Re s > 0 and Rev>0. Another representation by a definite integral can be obtained from the definition (1) and Plana’s summation ‘formula 1.9(11)2B SPECIAL FUNCTIONS Lu () Oz, sv) = hots [los nme ste = 2S sinks log zs tan”! (/v) } (0? + £2)" (e?*~ I" de Rev>0. For z = 1 we have again Hermite’s formula 1. 10(7). Lipschit2’s formula 21s) z 28 (y+ nF = [Ee ! em (e — 6° (cosh t cos 6)"' de Pee. g6 Hee hee ee results from (3) by taking z = e ®, ® can be represented as a contour integral P=) [OP Gaet et (1 ze de Rev>0,_ larg(-2)| <7 (5) 2mi Oz, 52 assuming, as in the analogous work of 1.6, that the contour does not enclose any of the points t = logz + 2nzi(n = 0, 1, 2,...), which are poles of the integrand of (5). Equation (5), for every fixed s which is not a positive integer, defines ® as an analytic function of z regular in the cut plane, and for every fixed z in the cut plane,® as an analytic func- tion of s regular, except possibly at the points s = 1, 2, 3, ...(it being understood that Re v > 0). As in the preceding section our function can be represented by a series. In order to do so, consider £ Caste (L- ze)" de over the contour C consisting of a circle K of radius (2N + Iz (W a posi- tive integer) and a loop L round the origin. The center of the circle in this case is the point t = log z(z # 1), and all points ¢ = log z + 2ni (n = 0, 1, 2,... ) are to be outside the loop, Letting N » ~, itis found that the integral over K tends to zero provided Re s <0 andOL ne Fence, we have @ FC pe vat] =(1- 2") Gs) = OCLs, D Re s>0, (3) -F (ans De - 29) Els) = 27* OL, 8,9) Res> 1. ae We therefore have the following integral expressions for ¢(s) [ ef, 1.10(3) and L11(3)}- (Ts) ls) = P87! (et= Ide = 28" ~#45"" esch t de Res>], (5) (1=2°74) M(s) Cs) cera (et+ Do! de = 2971 fe tee! seche de Res >0y ° : (©) 2T(s) (L= 27) &(s) = f° 497! esch ede Res>1. From 1.11(1) and 1.11(3) we have () s)= (541, 0)=[2°°'/P(s + DI f° t8 (esch 2)? de Res>1, (8) (1-2!) Es) = © 1, 6+ b= L271 Ms + DI SS e8(seche)* de Res>-1L The following representations of ¢(s) by means of contour integrals (9) ami &ls)=-TU~ 5) LO ae et Dh de (10) 2ni(1 ~2°7*) C(s) =-P(L=s) f° Caet (ete 1"! de (11) 4ni(L ~ 278) £(s) =P ~s) f°") 2) esch edtLiz GAMMA FUNCTION a3 with s#1,2,3,00, larg(-dl Lu L+ 2) + y] a!" * de, T(s) si = Tle ea 7V(n+9) Jo a Om+s)=(-)""' These formulas are valid for 0 < Res < 1 and y™ is defined in 1.16(1). Furthermore os) =(s- D's sin(7s) { tw'l4+ ad -(L4 da 8 de a(s ~ 1) lo Ce eee ee pe Finally we prove Riemann’s representation of ¢(s), (16) a-#776/2) Cs) = — + f (Oo ae e-0" |34 SPECIAL FUNCTIONS LR where ol = Ee’. ¥ 10,00, it)- 1, 6, being the elliptic theta function, The integral in (16) represents an analytic function of s for all values of s. From L.1(5) we have een dear PG /2 n° Re s>0, Hence we obtain w*71%5/2) Es) =f wl) 27°" de = J o() 67 de+ Jrod Cot dE But by means of Jacobi’s imaginary transformation of the theta functions (Whittaker and Watson, 1927, § 21.51) we have o() =- K+ ket se wa). Introducing this expression into the integral, we obtain w 2D /2) Ls) n= Us) + Mle D +f! oto FHM des fold PO de, and substituting |/t=t'in the first integral, we obtain (16). For further integral representations see Ramanujan, 1927, p. 72; Hardy, 1949, pp. 333,337. A power series expansion of ¢(s) is (Hardy, 1912, p. 215; Kluyver, 1927, p. 185) (17) ls) =(s- D7 4y+ 5 y,(s— 0 where eek. Yqm Tin CS Tog Dh = to Dog. ZA Putting v = Lin L10(8) to 1.10(11 we obtain (18) €@)=-%, $0) =~ % log (22); (19) Jim, [g(s) - Ms - D1 =- =, and (ef. 1.13(7)] B (20) om) =~ st m= 1, 2,3, 6065 m+LB GAMMA FUNCTION 35, or B. 2m) = =(-pe*! yen 2a (21) -2m) =0, (2m) =(-1)**! (2a¥ 2m) | m= 1,2, 8, 0065 B (2) ¢[-(2m- D)-- 4%, 2m Putting » = 1 in Hurwitz’ equation, we obtain Riemann’s functional equation for £3) (23) le) - cae sin(rs/2) Gs) or in view of L216) (8) 1s) =(20)"# BPC) cosas /2) Ll). Introducing a new function defined by s(s~ 1) 2 (25) &(s) = TUs/2) #**e(s) we have (26) €U- 8) = ls). This function is known as Riemann’s £ function. For asymptotic repre- sentations ofthe zeta function see Hutchinson, 1925; Titchmarsh, 1935, 1936; for numerous other results, Titchmarsh, 1930. If we consider the function co 27) L = s — Re s> 0, OD LO= 2. Gaede . which is similar to the ¢ function, we have by means of L11(1) nd 1.11(3) L 2 fe di = 27° OR —_—_ eee Res>0. (28) L(s)=27F OCLs, 9 ao f — es Putting z = -1, v = ¥4 in Lerch’s transformation L11(7), the fol- lowing functional equation for L (s) is found: (29) LOL ) T(s) sin (ws /2) L (s), n (For further discussions see Lichtenbaum, 1931, p. 641) 1.13. Bemoulli’s numbers and polynomials The Bemoulli numbers B, are defined by the equation QM) zfet-'= EB fal z < Qn, n=o :36 SPECIAL FUNCTIONS 113 and the Bemoulli polynomials B, (x) by means of (2) sem (et—Yt= Facey sal || < 2m Since the left-hand side of (2) is EB z/r! }. { g arya } ro azo Cauchy’s mule for multiplying power series gives (3) Badass (7) Barts vne( : Vane ( "Ya 1 n-1 nj a (7 aes : B=1, B(z=x—% B(x) = Pat 1/6, B,(x) = x°-3/2x? + ox, B(x) = x*~ 2x9 + 27- 1/30,.... Clearly we have (4) B,(0)=B,. Differentiating (2) with respect to x and comparing coefficients we obtain (5) Bi G)=nB,_, From (2) it follows that 2, (8, («+ 0-8, Hence we have By(x+ = B(x), By(e+ )~B, (x)= 1, and in general 1 (© Byle+D~B,()=nx"", oe from which it follows that () B,()=B,0) =B,. n>2 Since we have BGs) i nt ie Cauchy’s rule for multiplying power series gives a recurrence formula113 GAMMA FUNCTION 37 for the Bernoulli polynomial: @ 3 (") Bx)=B (e+), or "S) @) Bix) = nx"! rEo\r orl O64: n From (5) and (6) we obtain Baer (Y)~ Big, (2) a > LO BG) deen", () f? BLO de = Hence it follows that Byay() ~ Bas +1 n=2,3, 4,.... From (6) we can obtain the multiplication theorem and the symmetry property of B, (x) (Fort pp. 32, 34) (10) "S'S" SB ate S" BO dee r=0 r=0 09 a QD B,(mx) =m! S B(x + 1/m), (12) BL (1-2) =(— 1" B,(). The Bemoulli polynomials are expressible in trigonometric series. For B, (x) we have from (3) (13) B(x) =%-% =- © (m7 sin(2arx) O 1 can easily be obtained by the calculus of residues, Consider f f(z) dz with f(z) = z7* e* (e7-1)7! (& an integer > 1), the contour C being a (large) circle with radius (2N + Dr (N an integer), center at the origin. The poles of the integrand are z,= Qnir, (r= 0, +1, +2, «+. ). The residues of the function f(z) forr = +1, +2, ... are easily found to be (2mir)"* e?"'*, and from (2) the residue at z = 0 is seen to be B, (x)/k !. The integral aroundthe circle € tends to zero as N » « provided 0 < x < 1, and by the theorem of residues we have By(x)/kl=— 3" (nin) * e? is, The prime indicates that the tem comesponding to r = 0 must be omit ted. This gives the expansions (n = 1, 2,3,...;0<%< D (19 B,,) = 2- 8°" (20)! E (2a) cos (2x2),38 SPECIAL FUNCTIONS 118 (15) Baa (2) = 2-D"*! (Qn + DY! ES (2m) *"" sin (Qarx). Putting x = 0 we get the following expressions for the Bernoulli numbers (cf. also Schwatt, 1932, p. 143): (16) B,, = 2(-1)"*" (2n)! (2ar)7 7" = 1,23, 0165 (12) Bye, =0 m= LAB. Equations (4) and (8) give a recurrence formula for the Bernoulli num bers (a) > (")a- n= 2,3, 4,008 From (18) and (3) we have (19) Bo=1, By=~-%, B,= 1/6, B,=~1/30, B,= 1/42.., Numerical values of the B,, up to By and recurrence relations can be found in Ramanujan, 1927, p.L. By using (14), (15), and 1,11(4), the following integral representa tions for the Bemnoulli polynomials are obtained: en * cos(2na)-e72" eo ee on f cosh (2t) ~ cos(2nx) O-Rexel sola 220-3 de a © sin(2nx) = eo of cosh(2e) cones) Oe eeee te In terms of Riemana’s zeta-function 1.12(1) we have (22) B,, = (- It" (2m)"™ 2(2n)! C(2n) n=0, 1,2, 0065 (23) B,, =~ 2n d-(2n - D1 a as is seen from (16) and 1.12(22). Fron: 1.12( 4) to 1.12(8) we find integral representations for the B. (n= 1,2, 3,644): (24) Byy=(-D™ 4n ieee (eae = (pt! 2a to e~"*esch(nt) de, 2nL131 GAMMA FUNCTION 39 (25) Be Dein 2) tee (eas Dard =(- Dt! an(1~ 2!) 1 £220"! 6-7 sech (ri) dt, (26) Ba, =(- I" 2n(2"— D~" fe! esch (70) ae, (27) B,, =(-1)"t! aw [> t?*Lesch (nt) ]? de, (28) By, = (-1)"*! (1 = 2172" P* f°?" [sech (at) 1?dt (For other results cf. Nielsen, 1923,and Ramanujan, 1927, p.1.) 1.13.1, The Bernoulli polynomials of higher order The Bernoulli numbers and polynomials of order m are defined respec- tively by (29) a, sea, 2 [le MD ee (eR ym . 2 Ba, + a.) 2%/n! |e] < 2rla|7 (30) a, sera, 28[(e 1) eee lem DI te = E BO ela, + a) 2a! i2\ < 2rla, n=o Here m is a positive integer, a,, ... , a, are arbitrary parameters, and ja |). (3) |ayl= max [lal For m = 1 and a, = 1, (29) and (30) reduce to (1) and (2) respectively. Clearly we have (32) BI? (Ola, +++ a,) = Ba, +++ a,), (33) BA) (ala,) = a? B (x/a,). From (29) and (30) (34 BOMsla, s+ a,)= > (7 steele, say). We denote (35) €=%la,+ +++ +a,) and (36) Di) = 27 B@) (ela, +++ a,).40 SPECIAL FUNCTIONS L131 It can be shown that (37) Din? =0 ned be We thus get from (30) (38) (@, +++ a2" [sinh(a, 2) ++ sink(a, 2)|"'= Edie) 2*/(20)! neo lal 0. a ae) i Using (14) we have -61, E, = 1385,.... An integral expression for E,, can be obtained by replacing L(2n + 1) in (16) by the expression 1,12(28), (18) E,, = CD (2/nyrt! [* 0 sech ede =(-D" 924! Sp em sech(xd) dt nO, Dee, ‘The Fourier expansions (12) and (13) can be replaced by integral ex- pressions. The result is:Lis GAMMA FUNCTION 43 (oy Bcc yea [Esa coskln [, cosh (27d) ~ eos na) 2=0,1,2.0, O 1 If we form the expressions (1 + z) + w(1- z) and G(1+ 2)+G(1-2) by means of (5) and (6), and take into account 1.8(7), 1.8(8), 1.7(10), and 1.7(L1) we obtain (1) w+ d= (22) = y= (a/D ctn(ae) — = CQn+ 2" Iz] <1, (8) G(L+ 2) =27' ~ zese(n2) + 20(D +2 = (1-27) 2n 4 22 lel <1. Using 1.7(1) we have from (7) sin 72 S& (22+) P42) =~ - eae (©) log ha+2) % tog ( n ) a i or, using the series l+z oo ci : , ae (=) x Qa+l we obtain 1 (10) log M+ 2) = % toe | = | oi es) sind T-z SB 1-f2n+D ay, —— 1-ye + 2 a. (1=y) « Fonmulas (9) and (10) are valid if |2| <1. Finally we give an expression of Mz) and yz) near z =~ m (m =0, 1, 2,... ). From 1.2(6) we have T(z) = 7 -D*AT- 2) sinfa(z + mI}. Expanding 1/I'(1 ~ 2) in a Taylor series near z =—m and using 1.13(36) we obtain QD (2) =(-D*/m IN (z+ m1 + Win + D + (z+ ml (778) + W?2(m + 1) ~w'(m+ D+ OL(z +m)? I}.LB GAMMA FUNCTION az Similarly from 1.7(11), 1.18(81), and 1.16(9) we have (12) YG) =~ (+m + Mn D+ ECE ded Beemer, 1.18. Asymptotic expansions In 1,9(5) we replace the expression within the parentheses under the integral sign by the right-hand side of 1.15(1). Since the conditions of Watson’s lemma are satisfied, we may integrate term by term and obtain the following asymptotic expansion (Stirling series) (1) log P(2) = (2 = %) log 2-2 + ¥ log(20) + x B,, /U(2n = 1) (Qn) 2°") 4 (27781) larg z| ¥)teg2 (97) | 14 2 ee -4) [ ic 12 ne sie (2) D(2)=e larg z| L Putting s = (n + 1) we obtain an asymptotic expansion for ¥'")(z) as given in 116(9). Finally we derive an asymptotic expression of log I"(z) due to Binet. In Binet’s first expression 1,9(4) we write the integrand in the form Yer * (et Do [et(e-2) +042) =% Enter 42)! (eV, eo replacing ¢' in the numerator on the left-hand side by its power series. Since, according to 1.10(3) me *(et- I de= P+) C41, 24D, we obtain (10) log F(z) = (z— ¥) log z- z+ % log(2n) a n +% 2 Gap nt he+d- This is Binet’s formula. A similar expression converging faster is Bumside’s formula (Wilton, 1922, p. 90) (1D log P(2) = (z= ¥) log (2 - 4) = 2 = 4 + Log (27) ~ Eelam, 2) /[2% 2n (20+) Rez> -K. From the left-hand side of (3) and (4) complete asymptotic expansions can be given, These are (12) log P(z +a) = (2 +a ~ ¥) log z- 2 + % log(2n) Oe : oa 2 Gap i a larg zl <7, n=, 2, 3, 0008Ly GAMMA FUNCTION 49 (is) P(e +a) Me +a)/Ie+ 6) Me +8) a,ta,- 2, ~B, ey 2 : pe * “Fel GrDG+9* larg 2| 0, real)50 SPECIAL FUNCTIONS Lp the absolute value of the integrand is comparable with (6) e~Kearttl I BvtAR-7 Pt py when |2| is large. There are four types of convergent integrals (1). First type: a > 0. The integral converges absolutely for |®| < an/2 and defines a function analytic in the sector |arg z| < min(7, a7/2). (The point z = 0 is tacitly excluded. ) Second type: a= 0, B £0. The integral (1) does not converge for complex z. For z > 0 it converges absolutely if y is so chosen that (7) -By>1+a; and there exists an analytic function of z, defined over larg z| < 7, whose values for positive z are given by (1). Third type: a = 8 = 0, A <-1, Here (7) is satisfied for arbitrary y. The integral converges absolutely for all positive z (but not for complex z) and represents a continuous function of z (0 < z <). There are nowtwo analytic functions, one regular in any domain contained in larg 2| < 7, |z| > p whose values for z > p are represented by (1), and another regular in any domain contained in |arg z| <7, 0 <|z| < p whose values for 0 p, There are two analytic func- tions of the same nature as in the preceding case. There is a discontinuity at z = p and the integral does not exist there, though it may have a prin- cipal value, The nature of the discontinuity, and the principal value, are discussed in the paper by Dixon and Ferrar. Multiple integrals of a similar structure occur occasionally. ‘An example for an integral of. the Mellin-Bames type is the following one (Whittaker-Watson, 1927, p. 289) (8) [Plat 9) P(E +s) Py~s) M(8~s) ds Ta+yTa+dOP(8+y16+8 = 2ni Ta+B+y+ The path of integration is indented so that the poles of I'(y —s) T'(5 ~ s) lie to the right and the poles of I'(a + s) (8 + s) to the left of it, and it supposed that a, 8, y, 5 are such that no pole of the first set coincides with any pole of the second set. [For further examples cf. 2.1(15) and section 7.3.6 and Ramanujan, 1927, p. 216.1 1.20. Power series of some trigonometric functions From 1.13(1) a number of trigonometric expansions can be deduced1,20 GAMMA FUNCTION (cf. also similar expansions obtained by Schwatt, 1932) such as (1) zeothz=22(e*-D'+2= 3 2B, 2V/(2n)! no = 2 E (yt! (an) 2 no 51 lzl Re b > 0, we have Euler’s formula (10) F (a,b; ¢3 2) = Pc) (P(b) P(e — B)I7* ff eh ~ EO ~ eed. Here the right-hand side is a one-valued analytic function of z within the domain larg (1 ~ 2)| < 73 therefore (10) gives also the analytic contin- uation of F (a,b; ¢3 2). To prove (10) for |z| < 1 we expand (1~éz)"* ina binomial series and integrate term by term; this leads to beta-integrals which can be evaluated by 1.5(1) to 1.5 (5).60 SPECIAL FUNCTIONS 213 From the identity (ng) GS 2-2) Sy sfe-(o+ b+ Del Sa a LAP NL ~~ ea) 8] wma Lb (= BL = ea) it follows that the right-hand side in (10) satisfies (1), and with s = -¢ that JO sta sO (1+ s2)"* ds is @ solution of (1) if Re b > 0, Re(a + 1 ~c) > 0, and larg 2| <7. With s = 7/(1~ 9) this becomes Sf (L= are tL = = 2) ede, and therefore (12) F(ab; a+b +1-e;1-2)=P(a+b+1-c) (PQ) P(a+1—e)]" x —. (L+ sO (+ sz) # ds is also a solution of the hypergeometric equation. Moreover, any integral Spee b= aerP (1 2)" 8 de is a solution of (1) if C is either closed on the Riemann surface of the integrand or terminates at zeros of t° (1 — s)°"' (1 = tz)"*"' Expanding (1 = tz)"* in.a binomial series and using the contour integrals 1.6(6) to 1.6(8) for the beta-function we find iD (ec) explin(b —c)] Te) eb) 2 sin vlc 5) x LOMB em 1 = te)" dt Reb>0, larg(l-a]Reb, larg(-a)| lif z+0. If we put z = 1, the right-hand side of (10) becomes a beta-integral and we obtain ffm 1.5(1) and 1.5 (5) (14) F(a,bs cs D=T) Mc a8) [P(e ~ a) Me ~8)]7" Rec> Red>0, Re(c-a-b)>0. We can show directly that (14) is valid only if c 4 0, ~1, -2, ..., and Re(c - a—5) > 0. From the recurrence relation (c -a)(c —b)z F (a,b; ¢ + 1; 2) =e [@c-a-b~ Iz-c+1] F(a,b;c; 2) +e(e~1)(1-2) F(a,b3¢ - 15 2) and from the remarks after (5) of section 2.1.1 we find that, for m = 1, 2, Soe (ce — a)(c ~ b) bre: De ate F(ab; ¢;\) aes 5) F(a,b;¢+1; 1) (c~a), (c=), = 8 Fabsctms @,¢ y, (a,b;c+m; 1) provided that Tim (1~ 2) F(@,b5 ¢5 2) =0 Re(e ~a-b)>0. If this is true ( as we will show presently), then we have for m+ c lim F(a,b;¢+m; l= 1, him 72a b= y _ Me) Me ~a~8) ane (@),(-a-), P(e-a) P(e -d) _ Me~a+mMe-b+m) ste Ple+mPe-a-b4m) and this together with 1.18(4) proves (14). Now (1-2) F@bye~ 2-14 2 (,-», x nt) 2" 4 0 for z+ 1, P+) PG+n) Pe-) ~ P(e+n-DP@+)) Pa) P() +0 for n>62 SPECIAL FUNCTIONS 2.1.3 This is true according to (5) if Re(c — a ~b) > 0. A second type of integral representations for the hypergeometric series is due to E, W. Barnes, (1908), who based the whole theory of the hyper- geometric function on the representation T(@) P(e) (25) To 8 ca - Mass) P(b +5) P(-s) : oa Te +s) i where |arg (-z| 1. 2.1.5. Quadratic and cubic transformations We may consider (17) to (23) as linear transformations of F (a,b; c; 2). If a, b, c are unrestricted then there exists no transformation of a higher onder (ise., @ transformation in which the variables are connected by a non-linear relation). If and only if the numbers t(l-c), +(a-4), 4(a+b-c) have the property that one of them equals % or that two of them are equal, then there exists a so-called quadratic transformation. The fundamental formulas are those of Gauss and Kummer: (24) F[a,b; 2b; 42/(L+2)7] =(1+ ai Floah bs b+; 27), (25) F(a,b; 1804; z) see este. =(1~2)7* Fla, %a+%—b; l-a+b;—~42/(1-2)*),215 HYPERGEOMETRIC FUNCTION 65 (26) F(a, a + ¥; b; 2) = 278[1 4 (1 24] * Fie, 2a-b+ 1b 1-U-)* 14-99% pte (27) Fla, b; a+b +4; 42(1-2)] =F (2a, 2b; 045 + %; 2). In (26), (1 ~ z)* is defined in such a way that it becomes positive if z is real and z < 1, A consequence of (27) and 2.10 (1) is (28) F (2a, 2b;a+b +; z+) ~ Pa+b+ Te F(a, bs % 2?) T(a+%) Tb +%) Pa+b+90e% ae : z Ta F@ F(a+%, b +4; 8/2; 2”), The series on both sides of (24) to (28) converge in a certain neigh- borhood of z=0, and the formulas are valid in the largest connected domain which is such that it contains the point z = 0 and that the series in the formula under consideration are convergent within this domain, For in- stance, (27) holds if |z| < (2% ~ 1)/2 but is not valid if z is real and 4% asecbe Ca, and here the coefficient of z” is & (Ya), (a+ ~b), 4)" (a+ 2r),_, Ae G+a-B, rl aan! Because of the relations 4°(4a), (Kat), = (a), (-D Cn), (n-r)l=al, BD ara, ,=(a+n, @),/(ay, and because of Saalschiitz’s formula (30), this is equal to i @), > (a+%-b), (a+n),(-n), — @, 0), nl 4, (+a-), a+, rl nl+a-5)* This completes the proof of (25). Applying (22) to the right-hand side of (25) we obtain (142)*F[¥a, 4a+ es L+a~ 6; 42/(1+ 2] =F, b; 1+ a-5; 2), and if we introduce 4z/(1 + z)? instead of z as a new variable, a relation equivalent to (26) is obtained. To prove (24), we show first that (32) F(a, b; 26; 42/(1+ 2)7] = (142) (1+ 299 Fike, bathe b+ Ms (22/(1 + 2°) 144, which is equivalent to (83) F (a, b; 265 z)=(1~% 2)" Fi Ma, ha +s b + Ms lz2-2)34, as may be seer hy introducing 4z/(1 + z)? as a new variable instead of z. From (26) it follows that the right-hand side in (32) is equal to (L+2)** F(a,a-b+% b+; 27), and therefore (24) follows from (26) and (32). To prove (32), we use (10)2.15 HYPERGEOMETRIC FUNCTION 67 which gives: (34) Fla, b; 2b; 4z/(1 + 2°] = F(25) (P(0)]7? (1+ 2 x fle" (14 22(1- 22) +27], Since the integral on the right-hand side remains unchanged if we write =z instead of 2 and 1~¢ instead of ¢, it is an even function of z, and by introducing 1 ~ 2 = cos 6 we see that the right-hand side in (34) is equal to 207th (1 + z8(1 4 27)" (28) [P(b)]-? x fF (sin 0)! [1 + 22 cos O/(1+ z*)]* d0. Expanding the brackets [ ] in a series of powers of cos 6 and evaluating the resulting beta-integrals according to 1.5 (19), we obtain 2-79 (1 + 2)8 (1 + 27) * P12) (PO) 1"? & LPOra+0@,, (22 \2 noo Tb+n+hy(Qn)! V1+2?/) ” and because of (a), = 2" (4a), a+), (2n) t= 2" 010), we find that this is (+2? ]* _ regres) [s2 ] 2 PO) +) 5 (4a), a+), ( 22 \m at G+mH,et \ Tea? Applying the multiplication theorem of the gamma-function to the factor in front of the sum, we find that this is the right-hand side of (32). If two of the parameters are unrestricted, there are only linear and quadratic transformations. A cubic transformation of the hypergeometric equation exists if and only if either 1-c=+(¢-6)=+(c-a~b) or if two of the numbers #(1-c), +(a-b), +(¢-a-8) are equal to one-third. For a proof of the main results, which are given in section 2,11, see E, Goursat (1881) and G. N. Watson (1909). There exist transformations of the fourth and sixth degrees where one of the three parameters is unrestricted (cf. Goursat, 1881, and section68 SPECIAL FUNCTIONS 215 2.11), Transformations of other degrees can exist only if a, 5, c are certain rational numbers; in these cases the solutions of the hypergeometric equation are algebraic functions (cf, section 2.7.2 and Goursat, 1938). 2.1.6. F (a,b; ¢; 2) as function of the parameters In maiy cases it is convenient to prove a relation (e.g., a linear trans- formation) for the hypergeometric series under certain restrictions (e.g., inequalities) for the parameters; for instance, it is easier to deduce (22) from (10) with the restriction Re ¢ > Re 4 > 0 than to use (13) for the proof of (22) without any restriction for the parameters, In this connection the method of analytic continuation (with respect to the parameters) is useful, Jt is trivial that F (a,b; ¢; 2,)/T'(c) is an entire analytic function of a, b,c, if z, is fixed and |z,| <1 since the hypergeometric series then converges uniformly in every finite domain of the (complex) a, 3, ¢ space. From (22) it follows that the same is true for all z, for which Re By 0.2.2.2 10 oy 12 4 -m m+1 c+m com=1 m+] m+1 m+1 =m-t -m-I-1 -m HYPERGEOMETRIC FUNCTION SOLUTIONS IN THE DEGENERATE CASE m+l+1 I+] ni, mi, mh ni, 141 I+] nei, ni Degenerate solution u rat. 2.9 (1) us, —-2.9(18) 4, 2.91) us -2.9(17) w yrat. 2.9() us 2.918) 4, rat. 2.9(1) 4, 2.9(18) 4, 2,9(18) u rat. 2.9(1) uy rat. 2.9(J) uy rat. 2.9(1) ugrat, 2,9(17) u, tat. 2.9(1) 7 Second solution us — 2.9(18) ae 2.91) @ 20(17) 4, 29(2) u, 2.9118) log. 2.3(2) u, 29D Jog. 2.3(2) us —-2,9(18) log. 2.3(4) u, 2.91) log. 2.3 (2) 4, 2.9(1) log. 2.1(18) ug deg. 2,9(18) uw, 2.9(18) log. 2.1(18) u, deg. 2.9(15) u, —-2,9(04) log. 2.1(18) 2.9(14) log. 2.118)SPECIAL FUNCTIONS 2.2.2 72 (zz “Bor nes (Ez “Bor mez tn (81) 6% ‘ws Se (es “BOT (63 *n (@a)6°z "Bop? (BL)T% “Bor «, (66% a (BI)Ts 80, sez * nopinjos Puosag (1)6'z "wah Iu [tu (sl) 6g"? wou Ztpburu @erzealn geuew ez ves!” I-t-u-ug— [tu (ezrwen vo etitu @ez i'n ‘yu i+w (1) 6% “eu u woninyos 4-2-0 2 ayesoue8og Teptu T+)+u otitueu L+i+utw pu ve (penupuo>) ASVD ALVYINAOAC AHL NI SNOLLNTOS [tu Ttw [tutu I+u Ltt T+ 1% 0 6t st an or SL asezHYPERGEOMETRIC FUNCTION 73 2.2.2 ZS “Boy (oz = *n (81) 6% “ws Sn Ss “Boy (81) 6S on (21) 6% “75 ss “Boy @)e% = fn HSS “Boy BU 6S Sn WHE “Boy (13) 6% on (21) 6'@ "wer Sn aor nos puoaeg, (et)6rs "18 n (egtea'n (21) 63 “wes 8n (Des twin (Des "wsS Hn (6G "wala ()eewiln (erg waln wonnjos arerous#oq gou~ tug Gtitw T+u+) L+u+ws7g Ttitmgtu wo su qup-9 [ou-u uap-w— T+u Sti+u Trt I+? I+ .+1 T+u Tau-u— @-7-u-w— unj—u=[— 8 uz 9% rd 8% a ser)4 SPECIAL FUNCTIONS 2.3 2.3. The full solution and asymptotic expansion in the general case 2.3.1. Limearly independent solutions of the hypergeometric equation in the non-degenerate case We may assume now that none ofthe numbers a,b, c — a, c ~b is an integer. Then two linearly independent solutions u,(z), u,(z) of 2.1(1) can be obtained from any not identically vanishing solution by analytic continuation along a path which encircles one of the points z = 0, », 1. If c is not an integer, we may choose (1) u,(2)=F(@,b; 632), u,(2)ez'"€ Fla-c4 1, b-0 +1; 2~05 2). Ifa-6 andc —a~—b are also non-integers, the analytic continuation of u,(z), u,(z) can be carried through by 2.10(1) to 2.10(6). If a b is an integer, but c is not an integer, formula 2.1(18) gives the analytic con- tinuation of u,(z) and u, (z) into the neighborhood of the point at z ==, and if c -a~b is an integer, we have fore =a+541,1=0, 1, 2,004 (2) F(a,b;a+b +1; z) TOPG+b+) 2 ©, Ta+PG+D .&, A-D,a! 7? i been es ero) ) (a+D, (+0, (b+), i [am log] 1-2" where k= b+ De yblat ls ~varn+D -Wb+nsl) and where ‘S\ denotes zero if] = 0. This result can be obtained from 2.10(1) by puttinge =a +b+l+e and by a limiting process ¢ + 0. In the same way we obtain for u, in this case (3) zt lPQQ— 6-1, L-a-1; 2-a-b=1; 2) PW PQ-e-b-D yyy WG 4-8), 0-a-d “PQ Ta-1) T(b-D n&> RiGaelt x [kh —log(1~ 2)] (= 2)" where kts VAL + 240) +9(L-+n) Was 0) — bn) and where ‘S' denotes zero if] = 0. n=0 The corresponding formula for u, is obtained by replacing a, b by 1+l-a 1+1-b respectively. If-c is an integer, we may take (5) u, (2) = F(a,; ¢3 2) c>0, (6) u,(2)=2'"* Fla-c+1 6-041; 2-e; 2) ©<0, (2) u,(2)= F(a,b; L+atb-c; 1-2) ltatb-c40,-1,-%..., (8) u,(2) =(1-2) 8? Flea, ¢ — 63 L-a-b+e;1-2) l+a+b-c=0,-1-2,..., The analytic continuation of u,(z) and u,(z) into the neighborhood of the points at z = 0, z= or z= 1 can be carried through by using the formulas 2,10(1)to 2.10(15) since we may assume throughout this section that none of the numbers a, b, c- a, c-b is an integer. 2.3.2, Asymptotic expansions The behavior of the solutions of the hypergeometric equation for76 SPECIAL FUNCTIONS 2a2 Jarge values of |z| can be fully described by means of the formulas for the analytic continuation into the neighborhood of the point at z =», Unless a - is an integer, every solution «(z) can be put into the fom (9) wl)=ay z+ A, b+ OC) HOLE) where A, and A, are constants; if a — b is an integer, z or z~4 has to be multiplied by a factor log z. The behavior of F (a,b; ¢; 2) for large values of |al, ||, |e| has been investigated by 0. Perron (1916-17) and by G, N. Watson (1918). If a, b and z are fixed numbers and |c| is large with the restriction larg cl < 7-6, € > 0, then for {2{ < 1 we have (10) Fabs 5 214 eee eee MAO e (©), a! With a slightly modified expression for the remainder term this remains valid even if |z| > 1, |arg (1— z)| < a, provided that Re c >=. Then, if Jc| +o and b is fixed, Re c > Re . For a sufficiently large value of n we also have Re(b + 1) > 0, We have ab @), ®, 2+ O(jo"™"}), CD) F (a,b; ¢5 2) 1-2 ++ pat = Pan (a,b; ¢; 2) e& yor = ECO Maen) 2/17) Pe- 8 Ta) nM] « x fb DOL as) UL = st) ds a, We split a=a+ia’,b=f+if',c=y+iy' into real and imaginary parts, Then, for 0 0. This proves (10) Lwith |p,,,| instead of O(|e|~*"') for a sufficiently large value of n], But then (10) is clearly also valid forn = 1, 2, ..., since each term of the asymptotic series in (10) behaves like |p,,,|. For more general results see T. M. MacRobert (1923) who has proved (10) for a range of arg ¢ which is > 7. If a, ¢ and z are fixed numbers, c #0, -1,-2,... and 0 <|2| < 1, and if |b| + © such that ~37/2 < arg zb < %n, then we have ce bz)" (13) F(a, 6; ¢; 2) = F(a,b; ¢; vy-| 2 or | [1+0( 8/791 ae n=o and here the asymptotic fomulas for the confluent hypergeometric func tion of a large argument (cf. Chapter 6 or Whittaker-Watson 1927, 16.3) give (14) F (a,b; €5 2) =e ™9[P()/T(c — )].{52)~ 2 [1 + O([b2|-")) +[P()/P(a)] e® (b2) [1 + O(|bz\7")]. Similarly, if - 4m < arg bz < 37/2, ne have (15) F (a,b; ¢; 2) =e!" *{ 1 (e)/E (e — a] (b2)~*[1 + O(|b2|~1)] +[I(O/T(a) e* (52) [1 + OC \bz|")1 In the case where more than one of the parameters tends to infinity, G. N. Watson (19 18) has obtained the following results. Let ébe defined by z +(z?- )* =e #8, and put 1-ef =(ef =) e* where the apper or lower sign is taken according as Im z % 0. Then for large | 4{ we have (16) (42-4 TA Flat dy a-c+ 14+ As a-b4+ 142d; 21-271] _ 2 Pa-b + 1420100 A elt NE T@-e+1+P(e-b+a) x (L= ef )-eth (14 oF) eve FHK [1 40( 479] where |arg 4} < 7-5, 8>0 and also (17) Fla+a, 6- A; e3%-%2) Ree) a a Gee Toles x NHL AA OE 4 eb imteK) oAM EIT. OIA")78 SPECIAL FUNCTIONS 2322 where the upper or lower sign is taken according as Im z 2 0 and where |Al is large, €= C+ in, ~ Wan w,+ 8 0, w, = tan™' (n/Q), - w, = tan" (9 - 2)/2) 120, w, = tan" [(7 + 2)/ C1, - w, = tan "(q/f) nso Here tan” ' x denotes the principal value of the function, i.e., ~ estan! 2 <4n Other cases where a, 6,¢ (or their moduli) are large have been inves tigated by M. J. Lighthill (1947), H. Seifert (1947), and T. M. Cherry (1950 a, b). The case where a = ipv, 6 =iv, c= 1, where g, v are real where p is fixed, and where v » » has been investigated by A. Sommer- feld (1939). 2.4, Integrals representing or involving hypergeometric functions Kuler’s integral 2,1(10) cannot be transformed into itself by an ele- mentary subsitution in such a way that the relation Fla, 6; ¢; 2) =F (b, a; ©; z) becomes evident, although this is a trivial property of the hypergeometric series. Wirtinger (1902) has given a triple integral for # which makes the symmetry with respect to a and 6 evident, and for the same purpose A. Erdélyi (1937 b) bas derived the double integral MoT T@TOTe-aTe-F x Gab eae ee tre) e de de (F(a, 65 63 2) = which can be obtained from 2,1{10) and 1.5(11). H, Bateman (1909) [cf. also A. Erdélyi (1937 a, b)| has proved that (2) F(a, b5 65 2) =(o)/LT(s) Pe - 5) 3 x flea ayer! Fd; s; xz)dx Rec>Res>0, z#1, larg(l-2)| 0, Re(b+s)>0. Splitting the right-hand side in (4) into the sum of two integrals extended from 0 to 1 and from 1 to « respectively, applying 2.1(17) to the inte- grand of the latter integral, and substituting - z and — z~' for z, we obtain: So7F (@ b5 05 = 2) 270d met Ff" Fa, by 05 2) xP de +P) TG -@) (PO) P(e -a)]t e #7 2) xf F(a, l-c+a;1-b+a;2)2¢t8"' dz +P(c) P@-b) P@ Pe — d)]t etimbts) x [' FQ, l-ce +b, L-a+b, 2) bt dz ° where throughout we may take either the upper or lower signs. If we now eliminate the third integral by combining the two formulas with the upper and lower sign we obtain, with s = w — %a: Dat w) Pa -u) Te -%a+w)T1-b+%a-w) : r@ Fy @,d-c+a, 1 “T(l+a-8)T@=a) &,0-b+a)n!l nt+w+ ha ,_ FO F @,O, 1 POTA-8,£,@), nl n-w+ka (5) Se) TP(l+a-b)T(e-a)° T@ Te) Pd -68) x gketent dey Sp Fla, b3 05 2) 24" de.80 SPECIAL FUNCTIONS 24 The first equation in (5) is valid if Re (a + 6 ~ ) < 1; the second equa- tion is valid if we also have Re (4a tw) > 0, Re (1-6+@)>0, and Re } < 1. The formula becomes particularly simple if c - a= 1~6. We have (6) S& (cos 6) (sin 8)” € 1222 do = 2rterty tenor LT (g — yoy) P(Qu4+ D/P ta-v4p)) x F(-2v,a-p-v;l+a-vep;-1) 4 tent d gin *¥NT(g— yy) P(Qv+ D/P +a-p+y)] x F(-2p,a—-p-; lea-p+v;-) Rep>-%, Rev>-% In particular, fa =v +p +1, 2p = % 2v = y, we find from 1,5(13) (with a= 1, b=-1, v =e) that (ee ats [ee (con) (onde ite ae a2 (P+ YG + D/T@+y 4D =QG+D' Fx 1l3y+25-D+ @+ D7 Fy, 15% +23-D Rex>-l, Rey>-L These formulas may be obtained by introducing e??8 as a new variable in the left-hand side of (6) and applying Euler’s integral representation 241(10) to it. We may deduce from (6) for Re p > — 4, Rev > -% (8) {ico 6)*# (sin 82” e%2? de eo aD(Qv+) T(l+pt+v-a)T1+v+a-p) x F(-2p,a-p-v3lta-p+v;-l) if we define (cos 9 ?# = e~##[ sin (0 — 4x) P# kas On In particular we have for 2u = m= 0, 1, 2,... andv=0 F(-m, -4m-%a;1-%m-%a;-1) = (-2" (m + a) ese(an) J," (cos O* cos af dé a#0,+1,23.... The quadratic transformation 2,1(24) combined with 2,1(10) gives2.5.1 HYPERGEOMETRIC FUNCTION 81 (9) F(a, a—b + %; b +% 2”) Te+h) “TOTO SJ (sin g)®' (1+ 22 cos gp +27)" dd ) Reb>0, |z|<1. Ifn = 0, 1, 2, «4+, we have (10) F(a, n+a3n+1; 2%) “al “ho! cosng (1-22 cos +22)"* dg Rea>0, |z|<1; this may be shown by expanding (1=2z cos + 22)" = (1 = ze ##)7* (1 — ze“i4)“# into the product (@), ¥ a > @), tol! sz=om! If we multiply these sums and collect the coefficients of e*#$", we ob- tain (10). We also have 7 278P(14 p) TIV+%B+%a-¥%yT14+¥B- Kat hv) xF ev %a-¥B-%v; 14+%B+%a-%v; a2/b?) = J 63 (cos OF (a2 e 4 b%e 19) dO etn, an ReB>-1, ||> Ia. This follows from an expansion of the integrand in a binomial series and term by term integration. If |a| > |b|, the corresponding formula may be obtained by substituting - a for a and — @ for @ The analytic continuation of the hypergeometric function in (11) into the domain |a/b| > 1 does not give the value of the integral on the right-hand side. 2.5, Miscellaneous results 2.5.1. A generating function If n = 0, 1, 2, ves, the polynomials F(—n, a +n; c; z) are the Jacobi polynomials (cf. Chapter 10 on orthogonal polynomials), For these we have the generating function Q) Ss) Fenatnye;z)/n! n=o wgnif SHs-1\" (S+s+1 Ve" 2Qsz 282 SPECIAL FUNCTIONS 25.1 where S=[1-2(2- 22) 5 + s?]* and $ + 1 ass +0, The expansion on the left-hand side in (1) is con- vergent if |s| < 1 and |1 - 22| <1. In the case Re ¢> 0 we can prove () by introducing u = £( ~¢2) (1 - 7! as a new variable in the integral 2,1(10) which gives T(c) T() = - : x ae Laut Leu-U\O" du ae u U=[1+ 2u(1- 22) +u°)*, Applying the inversion formula of the Mellin transformation we obtain l-u+ ee @) Ge ). ra east ye af Faro 5 MORI /B~ iso Te) where 8 is a suitably chosen real number. The integrand on the right- hand side of (3) has poles at b =~n (n =0, 1, 2, «..). An application of the theorem of residues can be justified by the results of section 2,3 and gives (1) with s =—u, By analytic continuation of both sides of (1) with respect to ¢ it can be shown that the restriction on ¢ can be dropped. We may consider (3) as a continuous linear generating function for F(b, a— b; ¢; z). For a bilinear generating function and for many related results see A. Exdélyi (1941). By the same method which has been used in proving (1) it can also be shown that (2) F(b, a~ b; ¢5 z)= C= see(1~ s+ set F 5h (o), Fen, a3 05 2V/nl Js) <1, fs(1-2| L [eeaa1 Since u ‘u\ fu-m = atm!/(a+m)! na +m, m n we also ] have s Ae [eYO] LOAN) aa OAD): and ad with (15) this gives oo vel u fo +1’ 0 FOQ-2hb-O/()) Tegucvine (15) and (16) are called “‘Lerch’s theorem’’, If we put z = ~ 1 in 2,1(25) and use 2,1 (14), we obtain 2--P(l+a—b) P(%) Td-)b +%a)T% +a) ee EXOOL @<* and if we put a =— 7m where m is an integer, we obtain $ oe (")(s m—u=1 2" T(-m+u) ya n= nJ\n /\ n Fare im) 04 — Yer) In a similar way, the linear and quadratic transformations of the hyper- geometric series can be used to sum in closed form sums involving binomial coefficients by one or several of the following processes: @) gi special values to z, Gi) putting a or b equal to a negative integer, (iii) comparing the coefficients of the same power of z in differ FG, 6; l+a—b;-D=254 HYPERCEOMETRIC FUNCTION 87 ent expressions for a hypergeometric series, e.g., in both sides of 2.1(22) or in an equation which is of a similar type. So we have that F(n, ~n; 1; 1) = 0 due to 2.1(14), and this may be written in the form © (-D'm+n-D! n!n!(m—n)! = Um. Another example is Saalschiitz’s formula 2,1(30). The truncated binomial series act 1G) z nso \" may be expressed by two hypergeometric series, viz., a F(1,-a; -2)-2 TG esa F(m~a, 1;m+1;~-2) or by ( . ) 2°" F(l-m, ly @-m4 23-271) m-1 afm-2,m-3,...,0. The incomplete beta function. In mathematical statistics, the func- tions Bi, a= Sper =a! de and I, 9 =B,(p, 9)/B, (p, 9) are found, B,(p, q) is called the incomplete beta function. We have B,@, = p' x? Fp, 1-gpt+h), Bye, 9 = Pp) Tg) /T(p + The following recurrence relations can be derived from 2.8(31) to 2.8(45) [For an application cf. T. A. Bancroft (1949)]: «Ip, 9-L(p+ 1 d+(U-a) (p+ 9-0 =0, (p+q-pd Lp, - 97, 9+ D-pA-DL@+1, q-D=0, gp, a+ D+ pl (p+ 4 d-@+ aL, 9 =0 2.5.4. A continued fraction Gauss (1812) [ cf. Gauss (1876)] has found continued fractions for the88 ‘SPECIAL FUNCTIONS 25.4 quotients of certain associated hypergeometric series. A typical example is F(a, b+ e+], 2) F(a, b; ¢; 2) where (a+n~V(ce~b+n-)) 4, = "(c+ 22 - Ylo+ 2n-2) (+n) (e-a+n) When we have b = 0, we obtain a continued fraction for F(a, 1; c; 2). 2.5.5. Special cases of the hypergeometric function In many cases the hypergeometric series reduces to an expansion of an elemenatry function. In the degenerate case at least one solution of 2.1 (1) always is an elementary function. Results of the type F(a, a+ Ys Ms 2)= (1m 28) 24 (La 2872 ee may be obtained from the quadratic a sey by a limiting process b > 0, Various other cases where F(a, 0; c; z) is an elementary function are listed in 2,8(4) to 2,8(17). All of them either may be veri- fied directly or may be deduced from the linear and quadratic transfoma- tions. Other classes of hypergeometric functions which have been particu- larly investigated or which occur in the theory of other functions are listed. in the table on the following page.26.1 HYPERGEOMETRIC FUNCTION SPECIAL HYPERGEOMETRIC FUNCTIONS Parameters a b,c Variable Two of the numbers 1-c, t(a~6), #(¢-a—b) are equal to each other or one of them equals + 4-Ma mn, n+ 2v; v4 (n=0, 13+.) ~hKz PS =n, at ny (n=0, 1,232.6) % % 1 z MMs 1 ee VL, 1/m, 1/n or zero (, mn=1, 2, 3,...) Quotient of two solutions of 21). One of the numbers a, bc-ae-b z is an integer. e~o=1 z 2.6. Riemann’s equation Name Legendre functions Gegenbauer polynomials Jacobi polynomials Complete elliptic integrals Inverse of automorphic functions Degenerate case Incomplete beta function 2.6.1. Reduction to the hypergeometric equation Chapter 4 22 2.5.3 For proofs of the theorems given in this section see E, G. C. Poole (1936). If a homogeneous linear differential equation of the second order has only three singularities and if these singularities are of the regular type (cf Poole 1936) then the equation may be written in the form: du ea a ———— (-2,)@-2) (@=z,)90 SPECIAL FUNCTIONS 26.1 Here a, a/, 2, are constants such that z, = 2,, 2,= 2, but z,4z, 42,4 2,, and (2) The singularities are at z = z, (n = 1, 2, 3); the constants a,, a! are called the exponents ieiaiae toz=z,, We admit the case where one of the singularities is at infinity, and the coefficients of du/dz and u in (1) are obtained by an appropriate limiting process. We shall call (1) Riemann’s equation. The name ‘‘Papperitz equation” is also common, The constants A, = a, ~ a’ are called the exponent differences. If none of them is an integer, (1) has two linearly independent solutions u, (2), «, (2) for the neighborhood of z =z, such that uy(z)=(e~2,)°" Ev =z)" n=0 oe vf (e- 2)" OF course, v,, vj may depend on z,, 22, 2, but not on z, and v, #0, vf £0. Tf one or several of the exponent differences are integers, thea one or several of the series (3) may involve logarithmic terms, Since we shail be able to reduce (1) to the hypergeometric equation, we may refer to the preceding section for the details of the logarithmic cases. By the symbol eee Goes a a ai af af we shall denote the complete set of solutions of (1). Them B. Riemann (1892) has shown LV fe-nY lees (6) ) ( Pda, a, a, 2 2 aoa at = @,+p a,-p-o 4,40 22.6.1 HYPERGEOMETRIC FUNCTION or and Co oe oe ee (©) Pda, a a, f>-=Pda, a, «2, = a al a’, aot ae where Ac+B 42, +B Qe cr and A, B, C, D are arbitrary constants such that AD ~ CB# 0. From (5) we see that the product of a solution of (1) with a factor (Say 2) also satisfies a Riemann equation. Of course we can permute the sub- scripts 1, 2, 3 in (5). If z, =o, then z~ z, in (5) must he replaced by unity. If z,, 2) 2, are given, we can always find four constants A, B, C, D satisfying AD - BC #0 such that ¢,, ¢,, 2, are any three arbitrarily assigned numbers, Therefore (6) shows that we can always transform (1) by means of a homographic substitution of the variable into a Riemann equation which has its singularities at the points ¢,, ¢,, ¢5- Combining (5) and (6) we obtain ) zy 22 aa 7 ba | (8) P a = Cie ee es 0 «© 1 (2-2, @,- x P 0 a,t+a,+a, 0 Aoetes aes) : (2-2) (@,~2,) a@j-a, a,+a’,+a, aj-a, 4 Now the hypergeometric equation 2,1(1) is a special case of Riemann’s equation. This may be expressed by saying that 0 i. (9) Py 0 a 0 z l-c b ec-a-b is the set of all solutions of the hypergeometric equation. Therefore (8) gives a reduction of the most general Riemann equation to the special92 SPECIAL FUNCTIONS 26.1 case 2,1(1). Since there are 24 different ways to transfomn the hypergeo- metric equation into itself this reduction is not uniquely determined. This may be seen as follows. The hypergeometric equation is character ized among Riemann’s equations by the fact that its singularities are at 0, , 1, that its exponent differences at these points are ]~ c, a—b, ¢ - a—b respectively, and that one of the exponents at z = 0 and at z= Lis zero. The six homographic substitutions G0) Gen Vee 0) et (ee ea! lead to the six possible pemutations of the singularities. If we multiply the solutions of a Riemann equation with singularities at z = 0, «, 1 by 2°(1 - 2)’, then we can always choose p, * in such a way that one of the exponents at z = 0 and at z = ] becomes zero. Since we have the choice between two expoaents at each of these points, we obtain 2 +2 + 6 = %4 transformations of 2.1(1) into itself. This leads to 24 solutions of the type 2? (1-2) F(a*, b*; 0%; 2*) where z* is one of the expressions in (10) and a*, b*, c* are linear functions of a, b, c. These solutions are called Kummer’s series;they are given in 2,9(1) to 2.9(24), These 24 solutions can be arranged in six ‘sets such that the four series belonging to each set represent the same function; the six resulting functions which, in general, are different from each other will be denoted by uy, uy, +++» Ug. Any three of these are connected by a linear relation with constant coefficients, the resulting 20 linear relations (valid in the half-plane Re z > 0) are given by 2,9(25) to 2,9(44). 2.6.2, Quadratic and cubic transformations The following relations are the source of the quadratic and cubic transformation of the hypergeometric series: 0 co 1 cate 80 1 (00200 go 2 Pe 2 A af af a, Bo fas ae , | where a,+af +a, +a% = %, and27.1 HYPERGEOMETRIC FUNCTION 93 O = 1 i oo (2) P60 0 eee ee V3 3 a4 a, as ce where a, + @% = 1/3 and w = exp (2174/3), Both of these were discovered by B. Riemann and investigated by E. Goursat (1881) [ cf. also E. W. Barnes (1908) and G. N. Watson (1909)]. There exist higher transformations in a limited number of cases where all the constants involved are rational numbers. For these see E. Goursat (1881), (1938). 2.7. Conformal representations [For this section cf. Goursat (1936), (1938)]. 2.7.1, Group of the hypergeometric equation If none of the exponent differences 1~c, b= a c~ a—b is an inte- ger, then u,(2) = F(a 0; ¢3 2) () u,(z)=2'"¢ F(a c+, b-c +1; 2-6; z) are two linearly independent solutions of 2,1(1) which are one-valued and regular within the domain |z —'4| <}4, but at each of the three points z=0, ©, 1, at least one of them will have a branch point. We shall con- sider the effect upon the two solutions (13) of z describing any closed or circuit, beginning and ending for example at z = 4 and enclosing one or several of the branch points. We can reduce this problem to the follqwing one. What will be the effect of two simple positive circuits C,,, and C.,y beginning and ending at z = 4 and enclosing z = 0 and z = 1? The effect of any loop can be reduced to the effect of a sequence of the loops Coeyr cy and Cy, Clye where the prime denotes the corresponding negative circuit, Now we can derive from 2.9(25) to 2.9(44) that u, and u, are affected by Ci, and C,, in the way indicated by the arrows. We have oa Gc unre 7m, u,> By, u,+ Bu, @ c u,> Byu, + B,u,94 SPECIAL FUNCTIONS 27.1 where Ba 1 deetnlenens) 218679) sin rd) iu sin (we) ” Po) P(e-1) P(c-a)P(c-8)T@)Ta)’ Qin eit lero) TQ-c)TQ-e) B,. = Bin ett (cna-b) : - PQ -a) PQ-s) Td +a-c) (145-6) sin w(c —@) sinz(c — b) Byy= 14 2ie in (ened? sin (ne) The proof of (2) is obvious. To prove (3) we start with the formulas Ccf 2.10(1)] Fla, 63 ¢; z)= dy, F(a, bs} a+b-e+1; 1-2) +A, (1-288 Fle =a, -b; e~a-8+1, 1-2), 2" F(a-etl, b~ c+]; 2-6; 2) =hy z'¢Fla-c+1 b-c+latb-c+1 1-2) ty (l= 2errt 2t"¢ Pia, 1-b; cab +1; 1-2) = dy Fla, bs a+b-c+1; 1-2) +A, (1-28? Flea, c~ 6; c- ab +]; 2-2) This gives Ya Wy m Ag hy = (gy Agy — Agg Agy) Fla, bs a+ b-c4+1; 1-2), yy ty Dy Hee Og dys “Ay ded dee x F(e-a, e-b; c—a-b+1;1-2), and from these equations we have Rap ty Aya tyr dep ty Aap Coy Yay hy — Ay, My > expl Qaz(e-a~b)] (A, u,-A,, 42). Here LW Te-a~d) _ TO Ma+-0) eee a neo) Ti@) Te)” _P@-9Me-e-s) | T2-dTMa+b-0 a TQ-aTd-)” ?) Ma-c+DTb-c+)"27.2 HYPERGEOMETRIC FUNCTION 95 The rest of the proof of (3) amounts to a repeated application ofthe fun- damental relations which hold for the gamma function, The effect upon u,, u,, of any closed circuit beginning and ending at a fixed point, say at z = 4, may be described by a linear substitution; the whole set of these linearsubstitutions forms a group, the group of the hypergeometric equation, All substitutions of the group may be obtained by a composition of those given by (2) and (3). Usually the set of homo- graphic substitutions for u ,(z)/u,(z) is also called the group of the hypergeometric equation. If ¢ is different from 0, +1, +2, ... , then (2) and (3) always have a meaning. Obviously, either u,, or u, is merely multiplied by a constant factor after any closed circuit if at least one of the numbers a b, ¢ ~ a, c — b, is an integer. This leads to the degen- erate case which has been investigated in section 2.2 If ¢ is an integer, it may be necessary to modify (2) and (3). To do this we may use 2.1(18), 2.1(14), 2.2(4), and also 2.10(7) to 2.10(15), For instance, in the non- degenerate case a = b = 4, c = 1, all of the three exponent differences l-c, b-a,c-a~—b equal zero, and in this case we may take {eons Ms 1s 2) 6PM, Ms Us 1-2 2 Then we obtain uyru, 4) Cy usr Quy,tus, u,ru,—2u, (3) Cyy Un rus. This result may be derived from (6) 4m F(%, 45 1; 2) + % log(L— 2) FM, 4% 15 1-2) -§ 0), nin! (y(n+)-v+%))] 1-2)" aa which follows from 2,1(15) by the method used in section 2.1.4. B. Riemann has shown [ cf. Poole (1936)] that associated hypergeo- metric equations have the samegroup of homographic substitutions and from this it can be deduced that there exists a linear relation between any three associated hypergeometric series where the coefficients are rational functions of the variable.96 SPECIAL FUNCTIONS 272 2.7.2. ‘Schwarz’s function (cf Kampéde Fériet (1937), Poole (1936)].. From now on we shall denote the exponent differences by (2) l-c=A, b-aep, c-a-bev. Ifu is a solution of the hypergeometric equation, then x (z) = 28 AY — ErKY (2) satisfies a (8) BEA re z)y=0 dz where 1-\? ey? 1-4? 42° “4-9? * aaa Defining the Schwarzian derivative \w, z} of a function by aa dw ,>w 3 [d?w dw \* Z dz3/ dz 2 a) ae and putting wle)=¥,(/¥(0 () TA, p, v3 2 = where y, andy, are two linearly independent solutions of (8), we have (10) tu, zb= 2A, p, v3 2). If Cis a function of z we have Cayley’s identity; ee (1D ty, 2h = te, oG) 4igszh dz Also a2 Ax +B Ax+B 0 where A, B, GD, are constants such that AD - BC # 0. Therefore, if = (Aw + B)/(Cw + D), we have Aw+B tm t= {EE “} Cu+D This shows that if w(z) is a solution of (10), then (Aw + 8)/(Cw + D) also satisfies (10), and it can be proved that all the solutions of (10) are of this type. Therefore we know all the solutions of (10) if we know two linearly independent solutions of (8). On the other hand, if w(z) is a solu- tion of (10), then w(z) cannot be a constant unless A* = np? =»? = 1, We27.2 HYPERGEOMETRIC FUNCTION 97 shall exclude this case, and then we have that dw \% di 4 (13) yt -0(22) vit) () are two linearly independent solutions of (8). We shall denote by s (u,v A; z) the whole set of solutions of (10), and we will call s the general Schwarz function. A particular Schwarz function will be denoted by Su, v, A; z). It may be shown that the func- tion s(A, x, ¥; 2) is meromorphic inthe neighborhood of any point z #0, w, 1, and that the correspondence between w and s is locally one to one (schlicht) since ds/dz or, if s has a pole, ds~'/dz is different from zero at all points z£ 0, , 1. This follows from 2G 2 ON ae vy dz Ye dz Ye where the numerator is a constant. If in particular A,,v are real, we may use the theorems on associ- ated hypergeometric series (cf. sections 2.1.2 and 2.7.1) to show that the functions s@tA,mtp,n tv; 2) i,m, n=0, +1, +2,... arise from hypergeometric functions with the same group if + m +n is even. Among these functions there is one set which we will call reduced and for which OSA H,v<1, OSut+y VHA, AtHSL Tt has been shown by H. A. Schwarz (1873) that a reduced function r= S(A, H, v5 2) maps the upper half-plane Im z > 0 upon a triangle A, in the mplane bounded by three circular arcs (some of which may be segments of straight lines) which do not overlap and which enclose in- terior angles Az, uz, vz, at the points corresponding to z = 0, 0, 1. Here the word interior refers to that domain bounded by the triangle which con- tains for example the point corresponding to z If follows from Schwarz’s principle of symmetry that the complete ‘system of branches of r= S(A, 2, v; 2) maps the z-plane upon a Riemann surface spread over the rplane which consists of A, and of all the tri- angles obtained from A, by the following construction. If a circle (or a straight line as a limiting case of a circle) is given, we can map the plane upon itself by a substitution aT +a, 47+ Ay where 7“ is the point corresponding to r and where ris the conjugate com-98 SPECIAL FUNCTIONS 27.2 plex value of 7, such that r= r’ for the points of the circle, We shall call that an inversion with respect to the circle. Carrying out the inversions with respect to the circles bounding A,, this will be mapped upon three new triangles A,, A,, A,, which again are bounded by arcs of circles, and we obtain new triangles from A,, A,, A,, by carrying out the inver- sions with respect to these boundaries and so on. If any two of the tri- angles thus obtained do not overlap, the function r= S(A, p, v3 2) has a meromorphic one-valued inverse function (4) z= 60,2,45 9 which is called an automorphic function. As a necessary condition for the existence of a function ¢ with this property we have that A, #,v, must be either zero or equal to the reciprocal values of integers. There are 15 reduced sets of values for A, », u, for which-s(A, yt, v; 2) is algebraic. They have been listed by H. A. Schwarz (1873). His list (in which Number | actually contains an infinite number of cases) is given in the table below. Here x, p, are non-negative integers and 2p 0. (cf. Whittaker- Watson 1927, sections 21.7, 22.3). 2.1.3. Uniformization If we introduce r by z=K%(7) (cf. 2.7.2) as a new variable, F(a, b; ¢; z) becomes a one-valued func- tion of 7 which is defined and regular in the half-planelm r> 0, Wirtinger (1902, 1903), has proved the fonnula, (15) 41 (8) P(e ~8) Fla, 85 05 x29] = #*1(0)14,(0, at” Je Ou, 9 de where 6,(u, 1) 7" CGE ees ® = 1. a [ae 5 | Ec 7] (uy 1) 7% [Ou 2) ] IB 6,(0, 2) 6,0, The functions 0,(u, r) (= 1, 2, 3, 4) are the four Jacobian theta func- tions (cf. Chapter 13). The integral in (15) must he replaced by a con- tour integral if the conditions Re c > Re 5 > 0 are not fulfilled. (cf. 2.143). 2.7.4. Zeros Let u(a, b; 6; 2) be a one-valued branch of a solution of 2.1(1) which is defined in the halfplane z > 0 with the possible exception of the100 SPECIAL FUNCTIONS 274 points Q, «, 1. Then the equation (16) ula, bs cs z)=A is satisfied only by a finite number of values of z if A is any given constant, This follows from the fact that the solution of 2.1(1) can be expanded in the neighborhood of the singular points 0, «, 1, in the way shown by 2,10(1) to 2,10(5), From these we can deduce that the he- havior of u at these points is determined by a single term of the type vj@ec,le-2,)°° or v,(2)=e,(e~2,)° log -2,) where z,=0, , Land z— 2g is to be replaced by 27! if z, =o This is to say that u/v, approaches a finite value not equal to zero if z > zg» Therefore (16) can have only a finite number of solutions in a suffi- ciently small neighborhood of 0, «, 1 (if we confine ourselves to the domain Im z > 0 and therefore to a one-valued branch of v,). In the remaining parts of the upper half-plane u is regular and therefore (16) can be satisfied there only at a finite number of points. In the case where a, 6, c, are real, the number of zeros [ive., the number of solutions of (16) with A = 0] of a one-valued branch of u(@, b; c; z) has been determined by Hurwitz (1907), Van Vleck (1901), (1902), and Herglotz (1917). The methods used by the latter authors are closely connected with the results given in 2,7(1) and 2,7(2). In the case of the polynomials us Fen, atn; ys2) n= 0,1, 2,008 where a, y are real and y > 0, a+ 1—y>0 all the zeros are real and lie in the interval 0 < z <1, Gee Chapter 10 on Jacobi polynomials.)28 HYPERGEOMETRIC FUNCTION 101 SECOND PART: FORMULAS 2.8. The hypergeometric series & (a, ©), & ©), a! (QD) F(a, 6; ¢; z)= c# 0,-1,-2 2605 (a =1 (2) (a), =P(a+n)/T(e) = a(a+ D++s(a+n-1) n=1,2, 8000, Ifc =—m—I where m, 1=0, 1, 2, ..., then we have (@) Fem, b,-m-f;2= $ PO Ho (=m), nt Some elementary functions which can be expressed by hypergeometric series follow (cf. sections 2.2.1 and 2.5.5). (4) (1+ 2%= F(-a, 6; b; 2) (5) 414 2k 4 WL = 24) = Fla, a+ 5 Me 2) (6) [4+ (= 2f/2]'-** = F(a, a5 2a; z) =(1-2)% Fla, a+; 2a; z) (7) (1-2)724(14 2 = F(2a, a+ 1; a z) The truncated binomial series follow: (8) ©) Apo ¢) roe @) 2" Fl-m, 1; a-m+1,-57) Po te a oS (ye FT Plt tna em ain (10) e7* = (2 cosh z)"* tanh z F[1+%a, 4 +%a; 1+; (cosh z)~*) (LD cos az = FI a, — % a; %; (sin z)?] = cos 2 FL% + Ka, % — Ka; Ms (sin z)*] = (cos z)* Fl- a, % — Ya; 4; ~ (tan z)?] (12) sin az =a sin z F[% +%a,% —% a; 3/2; (ain z)?] =a'sin z cosz F[1+%a, 1-% a; 3/2; (sin z)?}102 SPECIAL FUNCTIONS 28 (18) si (14) tan™ z F(%, Ys 8/25 2”) z= 2 F(%, 1; 3/2; -2*) (15) log(z + 1) =z F(1, 1; 25-2) (18) log = = 22 FU, 1; 8/25 22) = a" ter A -1 been A Sy Leta] a0, 21 BEM FE, by 2 ELEMENTARY RELATIONS (18) F(a, b; ¢; 2) = F(b, a; ¢; 2) (19) lim [P(e)]7! FG, 85 ¢5 z) eon 3 = yer Onss sot Pagan sl, bent Lyn +25 2) @eD! (2) Fla 05 2)= oe Mote be) a [zete"! F(a, b; 0; z)) (21 (a), 297! Fla+n, b; ¢; z)= In de® - (22) (=n), 2° 1"" Fla by e~ 0; 2) = La" Fla, by os DI Z (23) (e = a), 28°21 (1 = 2) 8-8" F(a =n, b; 03 2) = [aicaeerey (1 ie) cree FCG, bio; =) dz” (c- a), (e~ 0), (0, (24) (1 etre" F(a, by +n; 2) [QL - 2) et" * F(a, b; ¢; 2)] dz” (25) COE Da (L-23°' F(atn, bye +75 2) es ee aga LA-2 Fla 85 65 2]28 HYPERGEOMETRIC FUNCTION 103 (26) (c =n), 2°71" (1— 2)" F(a—n, b; ¢~n5 2) = eet (1 abet F(a, 85 65 2] dz” (21) (e =n), 26-17" (1 = 2)8* 6" Flan, b—n5 6 =n; 2) a = sy let (1-2 *' F(a, Bsc; 2)). dz” ‘The relations between contiguous hypergeometric series in the case where two parameters are constants are given below: (28) (c -a) F(a~1, b; ¢; 2) +(2a-c— az + bz) F(a, b; ¢; z) +a(z-1) F(a+1, 6; ¢;2)=0 (29) (cb) F(a, b- 1; ¢; 2) +(2b-c- bz + az) F(a, b; ¢; z) +5(z- 1) F(a, b +1; ¢; 2)=0 (30) e(e- D(z - 1) F(a, bs e - 1; z)+ ele~ 1-(2e-2-6-D 2] x F(a, b; ¢; z)+(c¢~ a) (c~ 3) z F(a, b; c+ 1; z)=0. The fifteen relations of Gauss between contiguous functions [F de- notes F(a, 5; c; z), and F(a +1), F(b + 1), and F(c + 1) stands for F(a +1, b; ¢; z), Fla, b +1, ¢; z), and F(a, b; ¢ £1; z) respectively}, (31) [e~2a-(b-@) 2] F+a(1-2z) F(a+ D-(c-@ F(a-)=0 (32) (b-a) F+aF(a+1)-4 F(b+ D=0 (38) (c-a-b) F+a(1~ 2 F(a+ ) -(c~b) F(b-) =0 (34) cla-(c~B) z] F - ac(1-2) F(a+ 1) +(c~a) (e-6) 2 F(c+ D=0 (35) (e-a-) F+aF(a+1)-(e~ I F(c- 1) =0 (36) (c-a~b) F-(e-@) F(a-1) +(01-2 F(b+ )=0 (37) (b- a) (1~ 2) F-(c-a) F(a-1) +(c~8) F(b-D~0 (38) c(1- 2) F-c Fla-l+(c-b) zF(c+=0 (39) [a~ 1—(e - b- 1) ] F + (c - a) F(a 1) ~(c-1) (1-2) F(e-1)=0 (40) [ce - 26+ (b-a@) 2] F+ 611-2) F(b+ 1) -(c-8) F(B-D) =0 (41) ¢[b-(c- a) 2] F-be(1~ 2) F(b +l) + (c—a){e-b)z F(c+1)=0 (42) (e-b- YD F+b F(b+)-(e-)F(e-Y=04 SPECIAL FUNCTIONS 28 2 ge se (43) e(1- 2) F~c Fb- D+ (ca) Fle + =0 es (44) [b~ 1-(e-a-)) AF + (c-b) F(b- D -(e-D(1-2) F(e-1)=0 (45) cle-1-(2e-a-b-1) AF + (c-a) (cb) 2 Fle+)) -el(e- (1-2) Fle-1) =0 ‘VALUES FOR SPECIAL z To) Mc ~a-d) T(c- a) P(e - 6) c#0,~L-2..., Rec>Re(a+d) P(1+e-5)T%) TU -6+%a) Th + %a) 1+a-5#0,-1,-2,... (48) (a+ 1) F(-a, 156+ 3-1 +(b+ 1 F(-b, a+ 2;-D T(a+2)T(b+ 2) Tla+b+2 (46) F(a, 6; ¢3 I= (47) F(a, b; 1+a~b;-)= 28 = gatets a, b#~2,-3,-4... (49) F(1, a; a+ 1;-1) = 2olw(% + 4a) - YKa)] T(a+b+%)T(%) T(a+%) Pb +4) a+b+%#40,~1, -2,... r@) TC) THa+%d) %b-%a+') b£0,-L-2 0. (50) F(2a,2b;a+b+%; %)= (5) F(a, 1- a; b; 4) = 2" (52) F(2a,26;a+6+1; %) = (a-d) Tat b+ DUIP(A Pb + KI '- (a+ wT) a+b+140,-1-2... (4/3) P(2a + 3/2) 1(3/2) P(2a + 4/3) 2a43/240,-1-2..- T(2a + 5/6) (1/2) T+ 2) PMa+5/6) 2a+5/6#0,-1,-2,... 2a (53) F(-a, -a +4; 2a+ 3/2; —- vo-(@) a0 (54) F(3a, 3a+ 1/2; 3a + 5/6; 1/9) = @)29 HYPERGEOMETRIC FUNCTION ws (55) F(a + 1/3, 3a; 2a + 2/3; e*7) TQa+ 2/3) ina o(h(sort) oe ee . T(a+ 1/3) Pa + 2/3) (2/3) = dre (56) F(a+ 1/3, 3a; 2a+ 2/3; e *"/*) = Qn ewBite g-K(aat 1) Ta + 2/3) Ma + 1/3) (a + 2/3) 1(2/3) 244 2/340,-1L-% 20. 2.9. Kummer’s series and the relations between them Kummer’s 24 solutions of the hypergeometric equation are given below: () u,=F(a b; 62) ( = (15 2)87 978 Fle ~ a, ¢ - 8; €3 2) (3) =(-2)7* Fla cbs ¢; z(z-D) () =Q-27* Fle~a, 6; 6 z/(z- D) (5) u,=F(a b;a+b+1~c; 1-2) (0) =2'"€ Fla+l-e,b+1-c,a+b+1-c; 1-2) () = 27? Fla, at+1~c;a4+b+1-0; 1-27) (8) =z F(b+1-c, bja+b4+1—c;1-27') (9) w= (2) 8 F(a at+l—o;at1—-b; 27!) (10) = (2)! (1-2) 8 F(1- 8, e—b; @4 1-5; 27") (Q) =(1- 278 Fla, c= 8;a41-3;0- 271] (12) =(-2)'"*(1- 2° Fla+ 1-e, 1-5; a4 1-8;(1-27'] (18) u, = (-2)7? F(b+ 1l-o, 6 b+ 1~a;27') (ig) = (2) (1-2) Fla, c— a5 b + 1-9; 2") (15) =(1~ 2)" F[b, c- a; + 1-a; (1-2) (16) =(-2)'7§ Q- 288! Flb+ 1~o, 1-a3 641-4; (1-27) (17) a, = 2'7© Fla+1—c, b+1~-¢; 2~e; 2) (ig) = zt"! (L-2)8 8"? F(1—@ 1~ 4; 2-6; 2) (19) =2'"€(1=2)°' Flat l-c, 1-8; 2-0; zMz-D)106 SPECIAL FUNCTIONS 29 (20), = 216 (1-2) F[b+ 1-6 1-5 2-6; 2(z- 0) (21) uy =(1- 2°85 Fle -a, ¢~ bse +1-a—b; 1-2) (22) = 2171-7" FL-@ 1-8; e+ 1-2-8; 1-2) (23) z (24) me (y— z)e-e-b P(e a 1-a;c+1-a—b; 1-27!) me(1— 2) 88 F(e—b, 1~b; e+ 1-a-b; 1-27!) a z Any three of the functions u,, ... , u, are connected by a linear re- lation with constant coefficients. This gives 20 relations, viz.,: »POMG+i-9 | TOrMe-9 2 temo | Loe Pa+1-dMe-8) T(a+ 1-5) - om ere HAMO+I- | _ MT e-a Geo) 6 no T+ 1-d)TMe-a ge imetine) eittati-e) resis Te-)FG-8a = rey re-8) te“) ge Ce) ae Tie+l-a-b “8 re! cei) POO PO) TG+l-0 “4 Pe+i-9M@ _ re+1-9Ta-s im(bt1-e) Bee areca ce T@tb+i-c T@-9 us ceimisrney E00 ra. Tia+1-b “3 Me-aTQ-') — MaMe-a inte=a) = QIKse: a Cees to 4 etna) ra-) Ta — ' Me+1-9T®) | Me+1-9Td-a imtet ted _ P@+1-9Ta-a eo Tte+bet-e ? TQ-0 . geintari-e) PO TA~a) Ce.a HYPERGEOMETRIC FUNCTION 107 ra-)Tc-@ TQ-8) P@+1-0) eee i T+1-e- © TQ-0 : Pe-a)T@+1-0 joo AC T+1- . ica FA-aMe-) | PU-a Ma+1-9 ae a ros) Us Te -8) P(a+1-0) (Ail Uae oid T(a+1-)) : P@)Te-a-s) | P@Ta+b-~¢ |e 21c) = aero. POrG-9 | TOTe-b “Te-ar® “** Te-bre@ Parb+1-9PG~d | Marssi-a 0-0 Piesiie OE eee eee | Ti@+l-oT@+1-0 '* Tare . : T(atb+1-cTb-a (GL) eitt-b) 4 eimti-b) + eimttna) (34) (85) u, = eu, gee erreda! eon Ta+b+1-9T(@-b) _iny oro TQ-9Pe+1-9) — P@TA-ATa+1-B ery 6) "s"TqTprMert-4 ““T@-ere-)T@ ° * Te +1-a-)T(e+b-dT@+1-) 5, G9) = ~Ta-bTe-)Fe+b+1-0 “ Tar) Pat t=) sets, T(a+1-0 P(@ : (a9) we EGE D EMEA | POTO-OTG HHA gent ee eee : “= Tq-a@Pd+b-0 |! TQ-dTe-alr® © 7 co 2 lettre Tesh 9 TOI age TO =a P(e-aTa+b+1-0) T(a+b~) M+ 1-4) “PO +1- 9TH eitte-a) yar ot Ne aw 108 SPECIAL FUNCTIONS 29 ap o P@r a Be-a-8 T2-) Ma+b- 9) ped Gag) Ga Rare °“Ta-aTd-) “?*P@rl-dr@+i-0 * (Cone 3. 2 ores i T(2-) M(a-) T(1- 6) P(a@+1-¢) lay. Tle+1- enn | . rd-@ PQ) 7 , Desde a-d) Ple-D Te-aTle-8) Pe+1-a-b) Mb- a) TQ-aTG-a *° Pc+1-a-8) P(a-8) He eeereceensyeeee a eemeemmnmen a Td) Te-b) ined e u, (44) ~imleb) os -imle~ad Me These relations hold for all values of a, b, ¢, for which the gamma factors of the numerators are finite and for all values of z for which the series involved converge and Im z> 0. The first eight of these relations connect three of the functions u, , &,, ..., ug which are not defined in the same domain. The last twelve express a function which is defined in one domain D by two other func- tions which are both defined in one and the same domain D’whereD = D+ These formulas can be used for analytic continuation. If Im z <0 the signs of the arguments of the exponential functions in (25) to (44) must be changed, 2.10, Analytic continuation The fundamental formulas for the analytic continuation of the hyper- geometric series are given next, General case is where 1-c, b—a, and c~ b —a are not integers. [Cf. also 2,9(1) to 249(4) and 2,9(33) to 29 (44).1 () F(a 6; 652)= A, Fla bsatb-e+1;1-2) +A, (1-2)? Fle-a c—b; e-a-b+1; 1-2) farg(1- 2)| 1, Rew> 1/2 2imss14 SPECIAL FUNCTIONS 211 (39) F[3a/2, (3a— 1)/2; a+ %j~2*] =(1+2)'"™ Fla@-1/8, a; 2a; 248427) (1+ 2)79]. E. Goursat (1881) gives an extensive list of cubic transformations. Of these, the rational ones only will be repeated here: (40) F (8a, 304%; 4a 4 2/3; 2) =(1- 92/8)" x Fla a+¥; a+ 5/6; ~2727(1- 2) (92-8)"4 (41) F(a, 3a+%; 2045/6; 2)=(1-92)7-%* x Fla, a+%; 2a+ 5/6; —272(1- 2)? (1-92)~7] (42) F(3q, a+ 1/6; 4a+ 2/3; 2) =(1- 2/4)" x Fla, a+ 1/8; 2a + 5/6; -2727(z -4)79] (48) F(8a, 1/3-a; 2a+ 5/6; z) =(1- 42)7* x Fla, a+1/3; 2045/6; 272(4z-1)73] (44) F(8a, 3-4; 1/2; z)=(1- 2)? x Fla, 1/6 — a; 1/2; (2/27) (9 ~8z)? (1~ )~'] (45) F(3a, a+ 1/6; 1/2 z)=(1-2)7* x Fla, 1/6 ~ a; 1/2; (2/2) (2-9)? (1- 2)7?] (46) F(3a + 1/2, 5/6 - a; 8/2; 2) =(1- 82/9) (1- 42/3)790-? x Fla+ 1/2, a+ 5/6; 3/2; z(9- 82)? (4z ~3)79] (47) F(3q, + 1/2, a +2/3; 3/2; z) =(1— 2/9) (1 + 2/3)"307¥? x Flat 1/2, a+ 5/6; 3/2; 2(z-9)2(z+3)79]. 2.12, Integrals : ff etacarssrea ° Rec>Reb>0, |arg(1-2)| <7 Te) (1) Fla, b3 ¢; )-THTelD (2) F(a, b; ¢; z) a IT (0) explin(d— 0] ee Se aera cece ecceeeae ee 1- “di WG) Me inate ad, a Reb>0, arg(l-2)| Reb, |arg(—z)| <7, b# 1, 2,3, +46212 HYPERGEOMETRIC FUNCTION 1B -T'(dexp(-i xe) (8) Fla, bi ek 2) = TET (eb) sinlad) sin we -D) (140415077 «f 2b gerb (1 = 22) de b, c-b#L, 2,3, oe (8) Fla b;¢;1-2) Be xf sO "L4 8 (14 82)" 8ds 7 “TOTe- Rec>Reb>0, largz|Rec>Reb, |arg(z—1)| <7 2 270 [ (sin ¢7! (cos ¢)2¢7 7" ee ayer cee cence EE M Fa bad-KaTeoH J, (ce sin®e : @) Fla bez)" fee a : Tb) Me -8) (1-4 z+%z cos t)* 2'-e P(e) i (sin 2¥¢726"! (1 cos #)7¢ . TO Te =o (1-%z+lz cos 2) seal ta 2e+t 5, yze—2b-t her 2T(c) (cosh v)* (sinh v)* Pb) Te- b) Jo U(cosh v)? ~ z]* (10) F(a, 63 ¢5 2) 2 P(o) f (sinh #P*7%*! (cosh ¢— DET ETE! 7 ——_—— dt TOTC-o (p24 % cosh : (11) F (a, 6; ¢5 2) 2h P(e) iE (sinh 2f47%"! (cosh + 1yetem2et! 0 == dt T@) Me-6) (h—2+% cosh a) i (12) F(a, 5; ¢; z) 20) 10 (sinh v)7" (cosh v)?#72*! hb gi armen POMe-H [(cosh v)?—z(sinh oy"116 SPECIAL FUNCTIONS iz (13) F(a, 8; ¢; 2) eo) = 1)" (cosh e+ 1)e7 e+! “TR Te =o {1+ 2+(1-2) cosh d]® (14) F (a, 5 ¢5 2) . _ 28 Te) i (sinb )2°7**" (cosh 2 — Dt “PO Pe =a)/o [4 2 +(1= 2cosh * dt (15) F(a bs 6; 2 oe oe af eH (Le eT (1 ce de The conditions for the validity of (7) to (15) are Rec>Reb>0. FEF For other integrals see SH45) 24(1) to 2.4(10), 2.1(34), 2.1(35) and 3,7 where the Legendre functions may be expressed by special hypergeometric functions. For integrals which lead to hypergeometric functions see also Chap. 7(Sonine-Schafheitlin and related integrals).HYPERGEOMETRIC FUNCTION uz REFERENCES Bailey, W. N., 1935 a: Generalized hypergeometric series, Cambridge. Bailey, W. N., 1935b: Proc, London Math Soc. (2) 38, 377-384. Bancroft, T, A., 1949: Ann, Math, Statistics, 20, 451-455. Barnes, E, W., 1907: Trans, Combridge Philos, Soc. 20, 253-279. Bames, E. W., 1908: Proc, London Math. Soc, (2) 6, 141-177. Bateman, Harry, 1909: Trans. Cambridge Philos. Soc. 21, 171-196. In par- ticular p. 184, Bromwich, T. J., 1947: An introduction to the theory of infinite series, Mac- millan & Co., Ltd., London, Burchnall, J. L. and T. W, Chaundy, 1940: Quart, J. Math Oxford Ser, 11, 249-270. Burchnall, J. L. and T. W, Chaundy, 1948: Proc, London Math, Soc. (2) 50, 56-74. Chaundy, T, W.: cf. Burchnall, J. L. Cherry, T, M., 1950a: Trans, Amer. Math. Soc. 68, 224-257. Cherry, T. M., 1950: Proc. Roy. Soc. London Ser. A 1202, 507-522, Dixon, A. L., 1905: Proc. London Math. Soc. (2), 3, 206-224. Elliot, E. B., 1904: Messenger of Math. 33, 31-40. Erdélyi, Arthur, 1937a: Quart, J, Math. Oxford Ser., 8, 200-213. Erdélyi, Arthur, 1937b: Quart, J. Math, Oxford Ser., 8, 267-277. Erdélyi, Arthur, 1939: Quart. J. Math. Oxford Ser, 10, 176-189. Erdélyi, Arthur, 1941: Proc. Roy. Soe. Edinburgh, Sect. A, 61, 61-70. Fricke, Robert and Felix Klein, 1897-1912: Vorlesungen iiber die Theorje der Automorphen Functionen. 2 Vol., B. G., Teubner, Leipzig. Gauss, C. F., 1876: Werke, Vol. Ill, Kénigliche Gesellschaft der Wissenshaften, GSttingen, Goursat, Edouard, 1681: Ann Sei, Eeole Norm. Sup. (2), 10, 3-142. Goursat, Edouard, 1884: Moth, Ann 24, 445-460. Goursat, Edouard, 1887: J. Math, Pures Appl. (4) 3, 255-305 Goursat, Edouard, 1888: Acta Soc, Sei, Fennicae, 15, 47-127. Goursat, Edouard, 1936: Propriétés gengrales de L’équation d’Euler et de Gauss, Uctualités scientifigues et industrielles 333), Paris. Goursat, Edouard, 1938: Integrales Algébriques. Probleme d’inversion, (Act= ualites scientifiques et industriélles 692). Paris, Herglotz, G. A., 1917: Ber. Sachs, Ges. Wissensch., Leipzig 69, 510-534. Heymann, Woldemar, 1899: Z. Math, u Physik, 44, 280-288.1B SPECIAL FUNCTIONS REFERENCES Hurwitz, Adolf, 1890: Nachr. Ges. Wiss., Goettingen, 557-564. Hurwitz, Adolf, 1891: Math, Ann, 38, 452-458. Hurwitz, Adolf, 1906: Nachr. Ges. Wiss, Goettingen, 275-217. Hurwitz, Adolf, 1907: Mack, Ann, 64, 517-560. Jacobi, C. G. J., 1859: J. Reine Angew, Math. 56, 144-165. Kampé de Fériet, M. J., 1937: La Fonction Hypergéométrique (Mémorial des Sciences mathematiques, Fascicule 85). Gauthier- Villars. Klein, Felix and Fricke, Robert, 1890-1892: Vorlesungen iiber die Theorie der elliptischen Modulfunctionen, 2 Vol., B. G. Teubner, Leipzig. Klein, Felix, 1933: Vorlesungen iiber die Hypergeometrische Function, B. G. Teubner, Leipzig. Lighthill, M. J., 1947: Proc. Roy. Soc. London, Ser. A, 191, 341-351. MacRobert, T. M., 1923: Proc. Edinburgh Math, Soc. Vol 42, 84-88. Mehlenbacher, L, E., 1938: Amer, J, Math. 60, 120-22, Meixner, Joseph, 1941: Deutsche Math. 6, 341-349. Mellin, H. J., 1895: Uber die fundamentale Wichtigheit des Satzes von Cauchy fir die Theorien der Gamma und der hypergeometrischen Functionen, Acta Soc. Sci, Fennicae, 20, No. 12. Mitra, C. S.,, 1943: J. Indian Math, Soc, (N. S.) 7, 102-109. Orr, W. Ma. F., 1899: Trans, Cambridge Philos, Soc. 17, 1-15. Peron, Oskar, 1916-1917: S.-B. Heidelberger Akad. Wiss. No. 9, 1916, No. 1, 17. Poole, E. G. C., 1936: Introduction to the theory of linear differential eque ations. Oxford. Riemann, Bernhard, 157(and 1892): Mathematische Werke (1892), 66-83, B. G. Teubner, Leipzig. Schwarz, H. A., 1873: J. Reine Angew. Math. 75, 292-335. Seifert, Herbert, 1947: Math, Ann, 120, 75-126. Snow, Chester, 1942: The hypergeometric and Legendre functions with appli- cations to integral equations of potential theory, National Bureau of Standards. Sommerfeld, Amold, 1939: Atombau und Spectrallinien, 2nd Vol., 800-806. Frederich Vieweg & Soka, Brunswick. Truesdell, C. A., 1948: An essay toward a unified theory of special functions based upon the functional equation OF(z, a)/dz = F(z, a + 1), Princeton University Press, Princeton, N. J. Van Vleck, E. B., 1902: Trans, Amer. Math. Soc, 3, 110-131. Watson, G. N., 1909: Quart, J. Math. 41, 70-79.HYPERGEOMETRIC FUNCTION 9 REFERENCES Watson, G. N., 1918: Trans. Cambridge Philos, Soc. 22, 277-308. Whittaker, E. T., and Watson, G. N., 1927: A course of modern analysis, 4th edition, Cambridge. Winston, M. B95: Math, Ann, 46, 159. Wirtinger, Wilhelm, 1902: Akad, Wiss. Wien. S.-B, Il a, III, 894-900. Wirtinger, Wilhelm, 1903: Akad. Wiss. Wien S.-B. Ma, 112, 1721-1733,CHAPTER Il LEGENDRE FUNCTIONS 3.1, Introduction ‘The expression (a? — 2ar cos y + r?)~* represents the potential at a point P of a source situated at A when r and a are the distances re- rrr PA at O. The expansion of this expression in ascending powers of r is of the form (@) (1/2) s. P,, (cos ) (r/a)" 0L "The functions P#(z) and Q(z) are known as the Legendre functions of the first and second kind, respectively. They are one-valued and regular in the z-plane supposed cut along the real axis from 1 to — ©. We assume (6) larg(z +D| <2, |argel< 0. From (1) it is found that the Wronskian WIP H(z), QU (2) }=PH(2) $ OH (2) = Q(z) - PH (2) must be of the form c/(1 — z*), where the constant c may be evaluated by putting z = 0. Using (22) and (40) we thus obtain (13) WEPH(2), (2) e #7 HI (1 + Mu + ov) (4 + Mu + hv) (=) P04 hv 4 PG ke Bn) | There follows a table of expansions of the form given by equations (9) and (10).124 (14) SPECIAL FUNCTIONS 3.2 EXPANSIONS FOR P‘(z) (2+ DM (2-1 /P =p) (15) (16) LJ Pep) (2+ DM (2- D/L (Le =p) Dev =p) 2 ant sin(un) Mu) (2 ~ D4 (z + YM ef 2 (24 WY (2 ~ 1/1) 0 =P (ap) ef A (2 4 1h (z — DEY «(Pd +y =p Py -w))" at #9 sin(vm) Py) eX OH) (2 — YEH (2 tt D1 = 2v) (2+ EH (2 H/T) Hy -p)] 2 P(1 + 2) (2 + DY (2 - H/T POO 4y wl3.2 (15) LEGENDRE FUNCTIONS lez by = b ey lev l-p -v l-p -v-p | _J (14), 2.10(6) 125 EXPANSIONS FOR EM(2) Remarks (14), 2.10(1) The upper or lower sign according as Im z <0 (15), 2.10(6) The upper of lower sign| according as Im z < 0 ev | levau]ite +1 ay | 22 /-—+——_}—_4 z-1 lev l+v-p]i-e lav | l+v—pl 2427 2 (as) ——}—__| » les | (16), 2.10(1) ~v -v-p | -2v126 SPECIAL FUNCTIONS 3.2 EXPANSIONS FOR 2/(2) 2 P(14 2) (24 DM (2 — DH Pe) PO ty pl (ag) 2e1 PD (-1 - 2p) (2 + DH (2 - DH P(-v) P(-v -) 1 (2? - 1 /P(1-p) (20) 0 298 eK Do mv) (2? - HY /P(-v 2) (21) 2 (K+) (2? - DY /P (14-0) at eX (22 D/L = ev — ew) DL + Ky ~ Mw (22) (23) ts aa HY (2? — NHL + hy - hn) (Ky ~ Ke] Bret pK (= ay) eV tH! (2? — IHD (Hv =p) PD rg ty) 27H (2? — HH /T (14 =p)3.2 LEGENDRE FUNCTIONS 127 EXPANSIONS FOR P¥*(z) (14), 2.10(2) (15), 2.11(2) W+hv—hy written Wy+hv—Me aise: (20), 2.10(2) hy + 4p kv Mp | Bake Me (20), 2.10(D) 8 8 n My — he v +3/B %akv Mu | Lhe — he (21), 2.10(6) (23) “hy hy Amv128 SPECIAL FUNCTIONS 3.2 EXPANSIONS FOR P/'(z) 2 (2? — 1-H 2*4/T(1~y) 7H of ef KH) (02 WY AT ~My ~My) P+ Ky ~ Kw (25) wah Quel gh enkuty-1) 5(42 _ kv XID Ky Ky) 104 + Ky — ep" (2m)~# (2? —D-* D(- 4 ~) [ze = (2? - DA/d (26) (Qa)7% (2? — W7¥ DG + Lz (2? = DEK /PL ev =p) w7® HD —v) (2? - DAHL 2 + (22 = DEITY /D (vp) (27) 7 HEC 4 v) (a? — DLs + (2? — HP H/C + v =p) 2H (22 — pM Lz + (2? — DY /T— p) (28) ; ]3.2 LEGENDRE FUNCTIONS EXPANSIONS FoR 2!(2) (24) (20), 2.10 (6) Rez>0 (25) * (24), 2,20(2) The upper or lower sign 9/2 | according as Im z 0 aaa | Yap |3/2+y tosy Ete" =" (23 ( Mey |__| (23), 2,41 (16) %-p |er-v Ltvep]v+3/2 z—(z*~1)% | (0) ope (26), 2410(6) ~vap| hee ia Yop | 1-2] (24), 2111 2(2?-1)* (23) } Rez>0130 SPECIAL FUNCTIONS 3.2 EXPANSIONS FOR P}*(2) 4, L (29) (30) (31) SS —————————— a% PHD +e) (2? - DM Lz (2? - DEH /Ty =p + D st A 2 242? — Dz - (2? — 4*#/TO-p) 0 (2n)"# PU, + v) oF UH) (2? — YK Lz + (2? — 4 x[P@-p+ DI (amy PEK =v) oF UH) (22 ™ [2 4 (2? - EH x(Pey =I 7B SEPA Mv) (2? — Lg (a? = DEL wp) EXPANSIONS FOR e7 7” Q(z) 3 4 4 (32) Pav +u) Pew (2-0 Ce + DH AAPL + -p) YP) (2 + WH (2 -131 3.2 LEGENDRE FUNCTIONS EXPANSIONS FOR P/(2) a, b ey g Remarks a, b, eC nv-p [np fl- me 22? -)# 2 28), 2.10(6) 09) Yap [Mp |%-v (28), 2.10(2) z+(2? - 1)* 30)| ——- 80) 3G? — DF The upper or lower sign A al 0 -v-p | %-n | %-v | (29), 2100) sn z+(z?-14 a-e2- 1 The h.g. series converge Liv-p| %-p | » +8/2 | nowhere in the cut plane EXPANSIONS FOR e~ #” Q4(2) (37), 2.10(2)132 SPECIAL FUNCTIONS 3.3 EXPANSIONS FOR e7 *47 Q(z) met (Lay +p) Pp) (2 + D8 (2 — D/P (1+ - Ker Py) (2 WMH (2 + YH FIV PG) (2 + Dee (zg - HH (34) _ PS? A$$ Amat 2 Pd ey +o) Gp) (2 + DY (2 - H/T (Lt p) YP (-p) Psy tp (2-H (2 H/T tv =p) OY Py) (2 + DH (2 — Bure! 2 P(L+v) PL +y +p) (z+ DRM (2 - 1H /P (2 + 2v) YT(L4r) Psy +p) (2 + DM (2 — DHT (2 + 20)3.2 LEGENDRE FUNCTIONS EXPANSIONS FOR e” ™” Q4(2) 133 Remarks ley (36), 2.10(2) The upper or lower sign according as Im z 20 (36), 2.20(D (36), 2.10(3) (41), 2.41(17) (36), 2.10(6)134 SPECIAL FUNCTIONS 3.2 EXPANSIONS FOR e7 #7 Q4(2) TH Dy) (2? - Yo 2H P(L4y +p) Pew (2? - DH /D(1 + yp) tv et Dla y ty) (2? ~ 1) /T(y 4 3/2) a BID 4 ey 4p) eb i Uren) (22 _ yoke x(P1L+%v - yy) |vt 7H BDL +e thu) et AM (22 _ yn xIP(% +4y —Ke lit 2 PL ay sp) tH (2? — DEP 4 3/2) QU DG) 2H (2? — yr DTH P (Lay ep) Pp) 2H (2? - H/T a3.2 LEGENDRE FUNCTIONS 135 EXPANSIONS FOR e7 “” Q(z) (39), 2.10(2) (38) ]1- 2? Rez>0 (41), 2.10(6) ; e oly (4D, -2.40(2) a (40) | 2? leky . The upper or lower sign according as Im z 0 (4) z (38), 2.10(6)136 +—}+— SPECIAL FUNCTIONS 3.2 EXPANSIONS FOR & “ QU(2) 7 UD + yy + hp) of OH (22 — HY yp) (45) (43) wh PCL +My + Mu) oF oO 2( 2? — DAK yep) i. —4 Man)* PUL ty +p) (2? - Y™ Lz (2? — EYE + 3/2 (44) | o# BP (L4y ep) (2? = DHL 2 + (2? — DEPTH + 3/2) (46) QED) (2? — 1) a + (2? - Yt xe (Lev +p) Dep) (2? - D8 [2 4 (2? - 4 H/T 4p) (4 tery) (2? - [2 ~ (at Yee Ser (Ley +p) Pp) (2? - DM Lz (2? - DIH/ Pty ~p)3.2 1(43) (44) LEGENDRE FUNCTIONS 137 EXPANSIONS FOR e7 #” Qi(2) c 3 c 4 Remarks (41), 2.10(3) wy | -%e -%n [+n [A 4 — ev | 4 ky -%a [eke [8/2 Y%m-p |v+3/2 aa ee ‘The upper or lower sign according as Im z 2'0 (47)| +h | leven |v +3/2 cee (44), 2410( gay 44), (6) -v-n | %—p | 1-2 | (44), 2.1003) 2(2- 1 (00) -vip | 4+n | 1+2n | Rez>O cv-p | 4n-p | 1-2e 14 Qe (44), 2.10(2)138 SPECIAL FUNCTIONS 3.2 EXPANSIONS FOR e7 #" Qt z) (48) Chak DOK +p) (2? — Do * a — (2? - DELEIY/P CL + -p) wD (Lay +p) P(-%-v) cos(un) (2? - 7%“ (49) OM a DOG 4 v) oF OHM (22 MHL a 4 (2? — KEY x[Pd+y-pl! ~ 2a cos(un) PL +y +p) P(-K ~ 2) ef Om x (at — MH La + (2? — DAI3.2 LEGENDRE FUNCTIONS 139 EXPANSIONS FOR e7 ##” Q#(2) Remarks (48) (44), 2.10 (2) (48), 2.10(6) (49) ‘The upper or lower sign according as Im z = 0 The h.g. series converge nowhere in the cut plane140 SPECIAL FUNCTIONS 3.2 3.3.1. Relationsbetween Legendre functions From 3.2(3) we have () PM (2) =PH_ (2). From 3.2(5) and 3.2(8) we have 2) e#™ Pv sy + DOM =e HTT — w+) OH (2. From 3.2(5) it follows by 3.2(23) that (3) QE (2) sintw +p) - 4, (2) sintay —pl: =e” cos(vn) PH (2). From 3.2 (32) and 3,2(3) we obtain z (4) Q4(2) siner) = Kar et? [Pse-neck - Po (2) ] : and hence cup P@ned | (8) PIP) = FEET (PE Ce) fm) sinlun) ON) Ta+ep+D OL S ‘i (8) PE (a) = EE PIM) +(2/n) oH sinus) OE (2), and thus, ifu = m(m=1, 2, 3,...), oo fee ea earn acy From (5) and (3) e #!™T(v-p+)) ee eres sin[r (v — 41 [Q4(2) - OH, (2)] or with 1.2(6) 9) QM, (2 - O82) = eB costum) Py +p +) Pu -v) PSH (2). From 3,2(15), 3.2(3), and 3.2(32) (10) P¥(=2) = ef #PH(2) -(2/n) 6H sin (vy + 1] Q4(2) or (LD QE(2) e #" sink ay + pl] = Ma Le 2" PH(2) — PH(— alu Hence, replacing 2 by ~2 (12) Q4(-2) =e #7 Q(z),3.3.2 LEGENDRE FUNCTIONS 14h In (10), (11), (12) the upper or lower sign is to be taken according as Im 220. = If we replace z by z(z? - I-*,y by- 4 ~% andy by-v~% in 3.2(3) and compare with 3,2(44) we obtain Whipple’s formula (13) Qe (2) =e (4m)# PO + u + D(z - D4 BM [a (2? - DI, Re z>0, which is equivalent to (1) D(-v =p) Pl (2) = ie Yan) (2? — OCP [a (2? - I *] Rez>0, As z varies from a point on the real axis at whichz > [to a point on the imaginary axis, z (z* — 1)~* varies from a point on the real axis to a point on the cut between 0 and 1. As the Legendre function of the second kind becomes discontinuous on the cut, we must introduce the restriction Rez>0. 3.3.2. Some further relations with hypergeometric series From 3,2(3) and (11) (15) Qe(2) e #7 sin [ay + 4)] Tp) =a [ee F(-v, 741; 1-3 4%~%z) 1 ay" F(-v, v+11- wona], z+1 and hence by means of (2) Ke (16) 2 Q4 (2) eH P(e =F + +4) G3) Pu~o = F(-v,v+ 1; 1+4;% +2) tan z-l VY A i a ~ Fem SY) vy eds Leas ed with the upper or lower sign chosen according as Im z2 0. From (6), 3.2 (3), and (16) Ke (17) 7 PQ +m) PH(2)

1 when z is on the real axis to the right of | or to the left of ~ 1. From 3.2 (32) we have for the same function (9) e7 #" QH @ 19) _ PC +y +p) 0p) 4” be 2 2rd +v-p) z+1 x F[-v, l+vjltus 4-2 Kw +%0P@) (3) Flav, Lt v3 1-05 (1 ~ ale = From (19), 3.2(32), and 3.2(3) (20) Q# (2, 1-) - e7 #7 Q4 (2) = rie #7 PY (2). Hence, by means of (18) and 3,2 (44) Tw + 3/2) (ge yh — EE pe (2 @n)¥ (F -Ik EE PEO z+(t~ »"| 2(z2— 1)4 24% aie" Ee (at — WHO FL ay, ps 9/2 4 : 2(z? - 1)* ale + (2? ~ 41% af By Wm w5 8/2 +53.47 LEGENDRE FUNCTIONS 143 Furthermore from (9), (20), and 3.2 (44) (22) Q# (2) sin[n(v +1 - = (4n)% Pip =v) (22 - )¥ & #7 9) cos(um) [z + (z? = D4] (2> 0% ; | dem aom iors ESTE] conta 27 a ~z+(z?@-1)* * Len(a? 14] ef +s Bons hos =e |} The expressions (21) and (22) are not of the type 3,2(9) and 3.2(10). 3.4: | Legendre functions on the cut In many applications of Legendre functions z = x, where -1 0]. In order to avoid ambi- ~o guity, it is usual to introduce slightly modified solutions of 3.2(1), These will be denoted by P#(x) and Q(x). (1) P(x) = Le 4H" E(x + £0) + em MH" P(x ~ 10), (2) Qe (a) = Kem HT Lem UT OH( x + 10) + 0 4” QU(x - 10)] wl m, (1) and (2) are valid, and the hypergeometric series involved are polynomials of degree n ~ m inz. (id) If x =m, m= 1, 2, 3, vee, and m>n, then P#(z)andPH (x) vanish identically. However Pv ~ +1) PH(2) and P(e “p+ 1) P(x) approach finite limits as u > m,v > n. It is customary to write P°(z) = P,,(z) ete. Often P,(2) and Q, (2) are called Legendre functions, P/#(z) and Q#(z) associated Legendre func- tions. From 3.2(7) we have (@) B)=F4y,- 1s 4-K2). If we differentiate (3) m times with respect to z (m= 1, 2, 3, «.. )and consider 2.1(7), 1.20(5), 3.2(7), and 3.3(7) it follows that (4) pa (2) =(2 ~ 2 SBE) m=1,2,8,00+, iz and hence, with 3.3(11), and 3.2(8) pe 6) 93 -Gt-p¥ m=1, 2,3, e006 From (4), (5), 3.4 (2) and 3.4 (5) : 2 2" P, (2) ©) FeG)=(-D" 2" — 2,3.6.1 LEGENDRE FUNCTIONS 149 a” Q, (x) dx®™ m=1,2,8,+0., —-l 0 by virtue of 3.2(5} 2 and the contribution of the circular arcs to i vanishes with their radius. 1 Thus we have 271 Q,@) = JS" 10,(v -£0)- Q,(v +i0)] (e =v)" do, “1154 SPECIAL FUNCTIONS 3.6.2 and since the expression in the braces is 7i P (v), we have Neumann’s integral representation (29) Q,(2) =% Li (e= 0)! Pv) dv =D" Q (2). Writing (29) in the form (80) Q,(2) =42,(2) £\(2— oy! do -% J" (ew) 1P Co) — PW) do, we see by comparison with (24) that (BL) Wy) =% [1 (e297 (2) ~P.2)] dv. Generalizations of Neumann’s formula (29) are(Gormley, 1934, p. 149) (32) Qu(2) =%e #7 (2? ~ DH? [094 (PHD) do, v+u=0, 1, 2,..., Rev >-1, z not on the real axis between — 1 and 1; and (Wrinch 1930, p. 1087) See etreta! P.(z)Q,(2) =% Sf" (z -v)"! Po) P, (v) de, nS m, z not on the real axis between — 1 and 1, n, m integers. For the Legendre polynomials we have the generating function Py P,(z) for || < min|z + (2? - )*| (33) (1-2hz +h?) ¥=. iv P,(z) for || > max| z +(z? - D*}. This may easily be found by expanding (33) in a series respectively of ascending and descending powers of &, and using (16). (For a generali- zation cf, section 3.15). Gn the other hand if z = cosh(u + i) (u, v real), EAC) 7 converges for |k| Rev >~-1, z not on the real axis between — 1 and, Similarly from 3.2 (45). and 2,12 (24) Q) PO -p+ DP +) He) = 0 ao PEP apt D(z? - DHA x fle + (2? =1)* cosh 17”! (sinh 1) de, (3) P@~n+ DEM + COcosh a) =e #774 24D +y+1)(sinh a) x JS (cosh a + sinh a cosh t)"”-#"' (sinh t)* dt, ° both formulas valid for Re(v tu +1) >0. Setting e* = cosh @ + sinh @ cosh ¢ in (3) and using 3.3(2) we obtain a ue i: (4) Ot (cosh a) = Gsm) exrisinh a? i 7 8% (cosho—coshaY#™ do y PU-w eo0 Heep 0 Bea Ke Furthermore it follows from 3.2(36) and 2,12 (8) that when z is not on the real axis between ~ 1 and 1, < L (5) Qe (2) = eig-yt ees - nH x ST (e+ 608 HY" (sin 3" de Rev>-2, Rew +n+2)>0. From 3.2 (28) and 2.12(8) or from (5) and 3.3(14) we have “Kon (2 — 1yrH (6) P#(2) = aD [2 +(2? = 1) cos ¥* (sin 7 de - “TG=) -p») Reu<156 SPECIAL FUNCTIONS 3.7 From (6) a% 2 (sinh a)“ Py - xf (sin 2)"* (cosh a + sinh a cos ¢)”*# de Ren< % (7) PXcosh a) = and hence with the substitution cosh a + sinha cost =e”, (sinh a)* 1% ~n) x ie (cosh @ — cosh v)“#~* coshl (v +) v] dv (8) P#(cosh a) = (4m) Re u< %, Another integral representation may be obtained by considering J e%** (cosh a — cosh v)~#-™ dv taken around the rectangle with vertices (tc, 0) and(+c, iz), and having indentations at the points (+ a, 0). Making ¢ + » and using (8) we obtain an integral formula for P#(cosh a); in this we change v into ~v — 1 and then add the two expres: sions using 3. ie The result is P . aoe you (sinh a f sin(un) cosbl(y +) tl. oe TO) “(cosh t = cosh a)#*¥ © sin(vz) coshl +) ¢] “a J, (cosh ¢ + cosh a)#*# Rep< %, Re(v+p+)>0, Re(u—v)>0. On substituting for Z, ,y (2) in 7.8(6) from 7.3(31) and changing the order of integration, it is found that (10) (2) = etn)" (2? — HA Pa 39) x1 [7 (2-008 I cosl +) El de = cos(um) S* (2 + cosh ye e @*#) tae} Reu>~¥%, Re(v+p+l>0 where z is not a point on the real axis between 1 and —. Hence, with 3.3 (9) we have Pu +het - D4 Pusu DPuHe) x [Cz + cosh )#™ coshl(y +) 4 de Re(u-v)>0, Re(u+v+D>0 (1) P5#(2) = % x)3.7 LEGENDRE FUNCTIONS 157 where z is not a point on the real axis between - 1 and-~. Applying Whipple’s transformation 3.3(13) to (11) we obtain Pw+D . : (12) Qe(z) =e Tf [a 4221) cosht] 7?" cosh(ue)de y Po-n+n J, Re(v tu) >-1, v#-1,-2, Applying 3.3(14) to (10) ren) : (8) PH) => » sl 2=(2?=1)% cost?” cos(ut) de wine) Lz +(z? - 1)* cosh ¢]” e#*de} ReWv+u)<0, Rez>0, Rev0. This equation may be written (15) PS(z)= mre / [z+ (2? = )* cos ¢]¥ ede 2nP +1) Rez>0. Hence, substituting t = ® — & we obtain Tw+m+)) [ag faeces aaa (16) P 3 (z) cos (mp) PED x Pte 4(z? — 1% cos (® - P)]” cos(m®) dd Rez>0, Twim+)) P®(z) sii po ce ee ore) x [let (e? — 1 cos ( ~ YY sin(m) dd Rez>0, In case w = 0, (16) may be extended to unrestricted values of m and Re v > ~— 1 (Erdélyi, 1941, p. 351). In case m = 0 a generalization of (16) is (Whittaker-Watson 1927, p- 329) (18) Pyle 2! (2? - D¥ (2! - 1) cos yf] =(2n) [*Lz + (2? - D cos(y -®)]” x (2! +(z® ~1)% cos ®] 1" dd Rez>0, Rez’ >0.158 SPECIAL FUNCTIONS 3.7 Other expressions for P#(2) may be derived from the previous results by means of 3.3 (1). By means of 3,4(5), 3.4(8), and 3.4(2) similar expressions for P(x) and Q4(x) may be deduced, From 3.4(8) and (2) with z = cos 6 we obtain (19) Py +4) Vv ~ 4 +1) PH(cos 6) = in 2 E(y +4 + 1) (sin O)# x [Jf° (cos @ + i sin @ cosh t)-Y-#™ (sinh 1) de — SF (cos 6 ~i sin 6 cosh t)"Y-#"* (sinh 1) de | Rew>-%, Rel tp+1)>0. And from 3.4(2) and (2) (20) Pet OPO -p+1) Q4(cos &) =7% 2-H PY +p +1) (sin BH xLJ-°(cos @ +i sin @ cosh t)"Y-4"" (sinh t)*dt + £° (cos @~i sin 8 cosh #)"¥-#"* (sinh 1) de]. Rep>~%, Reluty+D>0. From 3.4(8) and (12), and 3.4(2) and (12) respectively, we have iP@+D 7T@-p+) X cosh (yt) de — e* ™ [(cas @ i sin 8 cosh t)"”-* (21) P#(cos 6) le" #" [* (cos +i sin @ cosh t)"”~* Xx cosh(ut) de] w+) Wwy-e+D x cosh(yt) dt + e%i4™ [* (cos Oi sin @ cosh t)-”-! (22) Q&(cos 6) = [ei ["(cos 9 +i sin 8 cosh x cosh(ut) de], both formulas being valid for Re(v W>- Ly #-1,-2,-3,.0-- From 3.4(5) and (6) (23) P#(cos 6) Kou (sin Qe rt See ea a Tu-» J, Rep< ly and from 3.4(5) and (13) (24) P¥(cos 6) ef te P(—v) Boag | f" (eos d-ising cosht)” cos(ut) dt3.7 LEGENDRE FUNCTIONS 159 + sin(ur) J (cos 0 +i sin @ cosh 1)” et de] 0<@0. From 3.4(5), (16), and (17) Tw+m+) a Pes) x ['" [cos @ +i sin cos( ~¥)]” cos(m®) db, (25) P53 (cos 6) cos(mp) = i® Po+med QP w+ D x L" [cos 6 +i sind cos( ~Y)]” sin(m®) dd = 0< O< n/2, (26) P8 (cos 4) sin(mp) =i" In case 7/2 <0 0, z not on the real axis between — 1 and 1, may easily be proved from 3.6(3) and 2,1(7). The integral {,* (¢ - 1)" (z — )#"' dt, which occurs in the proof, may be evaluated by the substitution teo(z-1) +1 and 1,5(1) as (2-04 T(n4 DPW n+ 1 40h In the same manner (80) PW) P74) = (= 224A [' Po (ext de Rew>0, by means of 3,6(3), 1.5(13), and 3.4(5).160 SPECIAL FUNCTIONS 3.7 Similarly we have (31) PQ) QB) = eH (a? HA JQ (=) de lz]>1, Rew>o, Rew-n+D>0, by means of 3.2(5), 1.5(2) and the substitution ¢ = vz +z. Furthermore from 3,2(45) and 2.12(15) (32) 704 —p) Qi (2) = e 7 2 nH (2? — WHF [z 4 (a? — WHIP x Let — 0° 9 [2 4 (2? I~ cet? — ede Rez 0, ‘The corresponding expressions for P#(z), Qu(cos 9), and P#(cos 6) follow by means of 3.3(9), 3.4(2), and 3.4 (8) respectively. The formula HPL = 2p) (2? - WHA PO-pPCp-v TO wD x La dee ee de Reitv)<0, Re(u-v)< 1, larg(z +D| <7 (33) PH(2) = may be proved by writing 1 + 2¢z +e? =(1+2?[1~2¢(142)7? Q-2)ly expanding the integrand in a series, integrating term by term,and using 1.5(12), 2.1(2), and 3.2(7). For representations of the Legendre functions as loop integrals see Hobson, 1931, pp. 183 -200, 236-243, 266. 3.8, Relations between contiguous Legendre functions The recurrence formulas for the Legendre functions may be derived by applying Gauss’ relations between contiguous hypergeometric fune- tions, 2.8, So, we have from 3.2 (14) and 2.8(30) (1) PH**(2) + 2 +1) 2(z? ~ # PETC 2) =@- wu + D PE(2) from 3.2 (28) and 2,8 (28) (2) (2v +I) 2PY(z) =~ +) BAC) + +) PY_, (2) er and from 3,2(24) and 3.2(4) (3) PH, (2) ~ BA (2) =- (2 + 2 (2? - I PEN), yes The following formulas may be derived from (1) to (3) GQ) = 24D PE fe) — +p) & e+ DPA) = (2p +1) (2? - 14 PASY2),3.8 LEGENDRE FUNCTIONS 161 (5) BE (2) - PH) =-( — p+ D (2? - DE PEG), (6) zPB(z) ~ BE (2) =~ +4) (2? - 1% PHY 2), (=p) zPH(2) - © +p) BY (2) = (2? ~ D* PHN), (8) @ ~e +1) PE Az) -( + + Dz PHC) = (2? - 1% PHN 2), Differentiating 3.2(7) and using 2.1(7) adP#(z) ~% pp Bz (9) A =v ewer DEAD ape "@)-aay PH). Eliminating P4"(z) by means of (6) it follows that dP (2) (10) (=) = =~ 4D PM) ~ + Dz PH(2) vz PH(2) —(v +p) PE, (2). It may easily be shown that the formulas (1) to (10) are valid for OE (2). With P#(x + i0) = e“¥iur P(x) we have the following recurrence relations for Legendre functions on the cut: (LL) P(x) 4 Qu + 1) (1 = 22) PAP) + - 4) @ ty + D PH(x)=0, (12) (Qu +1) x PEG) = —n+D PH (2) +0 +H) PY, (2), v (13) PH_, (2) — PA, (x) =(2v +1) (1-27)* PE), (4) © -W) O=1 + DPA HW) He 4D PH) = (2v +1) (1-2?)* PH), (15) PH, (x) - xPH(a) =v — + 2 (1-2)* PE"(x), (16) x P(x) — PH, (x) =(v + u) (1 22)4 PEN(2), (17) & = px PH(a) —(@ +p) PHL, (x) = (1 x7)* PEC), (18) @ =p +1) PH, (2) -( ty + Dx P(x) = (1 x2)* PH 2), APH (x) ne (19) (1 = x?) + DxPH(s) (p+ D PH, (x) aovx PE(x) +(v +4) PY_, (x).162 SPECIAL FUNCTIONS 3.8 Here again it can be shown, that the formulas (11) to (19) are valid for the QM(x). From (2) we have Christoffel’s first and second summation formula (20) (G-2) £ (am +1) P(e) PLO) _ =(n + VP gy (C) B,(2) =P, 2) Py (2) @D @-2) z (2m +1) P, (2) Q, (6) 7 + DIP, (2) O12) -P,@) Qa (OO 3.9.1, Asymptotic expansions For large positive Re c, the hypergeometric series F (a, 6; ¢; 2) is an asymptotic expansion in c, even when the series does not converge, provided only that z is not a real number > 1. Hence, for fixed z and v, and Re y + ~,3.3(17), 3.3 (16), 3.2(3), and 3.3 (15) are asymptotic expan- sions of, respectively, P#(z), Q(z), P=z), and Q7M(2). The first, second, and fourth expansions hold for all z save for points on the real axis between — oo and — 1 and + = and + 1, and 3.2(3) holds for all z not on the real axis between ~ oc and-1, For fixed z and yz, and Re v + «, 3.3(21), 3.2 (44), 3.3(21) together with 3.3(1) and 3,3(22) are asymptotic expansions of respectively Pi{z), Q(z), PH, (z), and Q4,(z). The first, third and fourth expansions “hold for all z save for points on the real axis between — se and— 1 and + «and + 1, and 3,2(44) holds for all z not on the real axis between ~ - and + 2. The expressions 3,5(5) and 3,5(6) are asymptotic expansions in v of P#(cos 6) and Qi(cos 4), respectively, valid ife< O 0. Thus we have Te sued i () Qlcos = Fe 37a) ee x beosl(y + 4)0 +7/4+ Your] + O"")], and rv ta+D = He eae i 4 2) Phloos # = Tray (i sin) x Leos +4) 0-1/4 + Bur) +O") ]y e0. For small values of @ see 3.5(10). Formula 3.5(9) gives an asymptotic expansion in p of P7(cos 6).163 LEGENDRE FUNCTIONS 3.9.2 3.9.2, Behavior of the Legendre functions near the singular points (1+ A) A= A= H— %)30] H— +44 aD DT CDT ant a/al% 1) (ath) 809 (MH) I a piv (44-4 ie wx @~ DAM + 4) Tay w(O>) CH Dd efi —D anid ce Yh- Au (A-? Y)Bol %— art-49I ah =F) (T+ + NT CYA a 2/2) YD (T+ u—ay yim uy(L~ 7) (U4! + 4) Tay CT) T/ gL = *) an uso) Suypeary eee O>ney O< toy sg oaw ae tere tge gaya ga o>ney Oaoy ney onoy O-', z not on the real axis between — 1 and 1. Furthermore from 3,7 (16) ¥ rw+D (4) Bly +2 .aS =[z+(z? - D* cost - $)]” Rez>0. P 3(z) cos [m(v ~9)]3.10 LEGENDRE FUNCTIONS 167 Hence, putting ¢ ~0, changing v into ~ v ~ 1 and using 3.3(1) and 1.2(3) Pu-m+) Te+) = [z+ (2? =) cose]?! Rez>0 (5) B(+2 2 .= P (2) cos(mv) Dougall’s expansion : € 1 1 Oe onl a) P“(cos 6) ooh 7 von vent ore ee may be proved as follows. We start with the formula sin(vz) ¥ «ay Jcos[(n +%)v] -a(cos 6”) » in O72 e = Ga ee cost (v +%9)0] (cos v — cos 0)"% do (sing \7> 72" . aa y = WK dg, x Tat cos[(v +4) 6] (cos ¢ — cos 6’) ip. Using (7) again and integrating term by term we find in(v7) (8) P5(cos 6) B-(cos 6") = 2 1 1 . x = (-1" ~——— } P#' (cos 6) P™,* (cos 6”) n=0 von vtnel . -7<046'0, Rez'>0, arg(z—DiRev>0, (8) SQ, (2) Q,() de =Uo-v) (c+ + IT los D yo + DI Re(o+v)>-1, o+v 4140, 404-1, -2,~38,..., (6) le (0, Gl? dx =(Qv + D7 bv 4D Rev>-%, DL) Py) Pa) dx = 29? [low v) (04 4 I" x t2sin (a) sin(xo) [bv +) oles Di @ sin(no~mv)} ot+v+1#40, in particular for v =n, o = m (n, m integers) (8) JS} PG) P, (2) dx =0, (9) Ll (P(t deen? (v4 4! [w? = 2 sina)? +0! + Diy (10) £} [P (al? dx =(n +4)" a ap J’ + (2) Q,2) de =lo-v) (ote + DI" xP + I) -Ylo+ D] [1 + cos(on) cos(vml — 4m sin(vz -o7)} o+v+1h0, v,0#-1,-2,~3,.005 (12) de [LQ GA dx = (2v + 1 Ya? - b' (v + D1 + (cos vz)?]} ' vé-1,-2,-3, 26, (3) J/, PL) Q,() dx =[e -0) & +o+ DI x [1 = cos(on - vz) - 277" sin (nv) cos(ro) [UW (v +) - Yo + Di Rev>0, Reo>0, c#v,3.12 LEGENDRE FUNCTIONS 71 (14) SL) BL) Q,@) de —n™! (2v 41)" sin(2um) yw + D Rev>0, (15) f' Py) P(x) de = 297 o-) (ot 4 DI xLA sin(4om) cos(Yyn) - A-" sin(4vn) cos(Hon)), (16) S32 Q, (2) Q.(2) de = (lov) (C4 + DIY +) - Wo }) ~ Yn (4 - A“) inborn Hun) ~(4 +A) sin(kow — Men) Rev>0, Reo>0, a) f' P@)Q.@de=[e-) (+ v4 DI" x[A7! cos (Myr - on) - 1] Reo>0, Rev>0. Tn (15) to (17) PH +h) PO+KA) D4 +%o)P 14%)" If in (1) p =p =m, v =n, o=1 (i, m, 2, positive integers) we ob- tain from 3.9(8), 3.9(10), and 3.4 (19) 1-(-p" (asm)! (=n)U+n+) (n—-ml Likewise from (1) ifm, n, 1, k, are non-negative integers we obtain (19) {'P A(x) PHC2) de =0 Ldn, A= (a) is Q%(x) PY(x) dx =(-D" 20) [PA PM) 1-22)" de =0 kém, 2D LY (PSM? de =n 7 (nt mI /(n~ mt, and (22) [= 22) [PR (a)? de =(n +m) Im (nm) Furthermore we have rh FIP aa) eee Regs 1 Td +ko-¥y bork s 7D (23) f'P, (x) a? dx = This can be proved by substituting 3.2(3) and 2.1(2) and integrating term by term. Using 1.5(1) and 2.1(23) we first find So PY dea (+) Fey y tot as) HOT M(o4 D' Flot y 42,0-v 4h ot 2; ¥) and with 2,8(50) the result (23) follows at once.172 SPECIAL FUNCTIONS 3.12 Barnes (1908, pp. 183 ff.) has proved that (24) Sf" 2? (1 = 22)? PH(x) dae . 2 Tw + Ko) P(1 +o) TO +ko- kv - 4 Pos % + 3/2 Rey~1, (25) (-)" 2847 Fm) fo x7 (1-24)? PR (x) de . PY +4 FO +k9TO+m+y TU +koskm—4v) V(3/2+koshm + hv) Reo>-1, ma positive integer, (26) {2 (1=24)"" [PH a)? de =- PO +p +)/(2uP =p +0) Reyu<0, v +p a positive integer, Other integrals involving Legendre and trigonometric functions are (Mac Robert, 1940, p. 95, 96; 1947, p. 366, 367): (27) (sin )°" PY (eos 0) dt TaD (ha + %w) Pha ~ Ku) “PO4 ha +4) Pa — 0) DOkn + hv + DP Gav +) Re(a ty) >0, (28) [> (sinh 1) °-" P>#(cosh 2) dé 2a + hu) Phy = a + DIK ~ Ka - hv) Te thy + DTG + he Bd PL +h uw — ha) Re(a+p)>0, Re(v-a+2)>0, Re(l-a-v)>0, (29) PCL + %v ~%w) 104 + By + Ka) Jf (sinh °°" G4 (cosh #) de =e HT Or sy + he) (1 + hv ~ Yea) Pha + WE ha—%y) Rela ty)>0, ReWv-a+2)>0. Furthermore (Shabde, 1945, p. 51) we have 0) (MO + DEG + DF Sf) Pyle) Pye) (L +a)" de =P ty + D/P (2v +2242) Relvtu+D>0. For further integrals involving Legendre functions see Chapter 7 and Bailey, 1931, p. 187; Banerjee, 1940, p. 25; Barnes, 1908, pp. 179 204; B. N. Bose, 1944, p. 125; S. K. Bose, 1946, p. 177; Dhar and Shabde, 1982, p. 177; Mac Robert, 1939, p. 203; 1940, p. 953 1947, p. 366, 367;3.13 LEGENDRE FUNCTIONS 173 Meijer, 1939, p. 930; Prasad, 1930, p. 33; Shabde, 1934, p. 41; Sircar, 1927, p. 244. For integrals with respect to their degree see MacRobert 1984, 1935, 3.13. The ring or toroidal functions The ring or toroidal functions arise when Laplace’s equation AV = 0 is transformed into toroidal coordinates (n, 0, ) Gy) xe £ Sih cos _¢ sinhy sing c sin @ cosy cen | «coal ens Oe coshy e088 With the substitution s = cosh y and V = (cosh 7 ~ cos 0) v(s, 0,6) the equation AV = 0 becomes ep 2, 2. @ 2 [tr-0 2] a oe os os “38 ag With v = 0,(s) v,(0) 0,(¢), this leads to the following differential equation for v, d*y dy ae () (1-8) St = 2s FY 9) +4) ~(1- 5°71 w?] 0, = 0, v and « being separation parameters. According to 3.2(1) solutions of (3) are PH (s) PH_, (cosh n) QP -y (s) OY _y (cosh n). The behavior for large values of 7 follows from 3.2(28) and 3.2 (45): (3) P(L=p) Py (cosh) = 2% (1 = e7@ )-# ev #80 x Fy -p, b+ —p; 1- 2p; 1-e77), (6) DL+y) QLy (cosh n) = 7% e #7 EO +v +n) (Le 7)H eH x FQtu, 4+v tp; Lev; e7), Special cases of (4) are (D Py (cosh 7) = C47 cosh 9/2)7* K (tanh 7/2), (8) Q_y (cosh n) = 2e"7/# K(e~7), (9) Py (cosh 2) = @/D7* 9 /* EL. =e") *]. K and E are the complete elliptic integrals of the first and second kind respectively (cf. also Darling, 1923; Lowry, 1926; Airey, 1935).174 SPECIAL FUNCTIONS 3,18 All other properties and representations of the ring functions follow from the formulas of the earlier sections of this chapter. (For an expan- sion theorem involving toroidal functions cf, Banerjee, 1938; 1942.) 3.14. The conical functions The differential equation d? di @ a3 ~ Be [pt a HCL 2 tw 0 (p is a real parameter) is a special case of 3.2(1) with vy =-% + ip. Solutions of (1) are 2) Phys gle) and yy glee The conical functions are solutions for a real argument x numerically less than unity. @) Py, g(cos6) and QHy, ,,(cos 6). The principal properties of these functions can be obtained from the general results regarding P/(cos 6) and Q4(cos 6). For instance from 3,5(7) and 3.5(8) 2 + (4)? (4) Py, pleos@) = 1+ a Oe ie +04) Ip? 28/94 pees in OS a eee 0< 0<\n, 4p? a1? ©) Py, gleos@) =1+ (sin 6/2? (4p? + 1) (4p? + 3°) oS Cain 0/04 + 0 0<0 0. (For further integral formulas cf. Watson, 1944, pp 367-369.)178 SPECIAL FUNCTIONS 3.15.2 3.15.2 Gegenbauer functions From 2s1(1) itis to be seen that (8) with n replaced by a (a arbitrary) is a solution of (21) (2? = 1) w" +(2v 4D zw! -a (a +2v) w=, We therefore define Gegenbauer’s function for arbitrary (possibly complex) values of a by (3) or (4) with m replaced by a. From (4), 3.7(6), 37 (8), and 3,7 (34), respectively (22) C22) = 9 Pa + 2) TO + W/O) PQ) P@ + DI xf" [2 +2? — DY cos ¢]¢(sin 0)" de Rev>0, (23) CX(cos $) = 2” 7 * Dla + WP +¥/IPO) P(2v) Pat DI x (sing)! f,* cosfly +a) v] (cos v~ cos g)”! do Rev>0, Oq+l. The numbers a,, ... , a, are called the numerator-parameters and Pyysseypq,are referred to as denominator-parameters. The senes Pu is not the only generalization of Gauss’ series. The hypergeometric equation is a linear differential equation of Fuchsian type, but, while the series (1) also satisfies a linear differential equa- tion, this is not of Fuchsian type. Thus the differential equation can be used ag the basis for another generalization in which one considers the generalization to be the solution of an equation of Fuchsian type of higher order. In this connection L. Pochhammer (1870) has investigated the most general homogeneous linear differential equation of the n-th order which has singularities ata,, a,,..., @,, © and is such that the 182 so), at41, GENERALIZED HYPERGEOMETRIC SERIES 183 general solution in the neighborhood of every singularity a, (v = 1, 2, +++ 47) is of the form 3, c,(2-a,)¥ + 2° 3, el(e-a)" where ¢,, ¢,, +++» Cy- and cf are arbitrary constants. Similarly, the expansion for large z of the general solution is assumed in the form 8 go vtaar 8 ges, » Enea and Bg are arbitary constants. It can be shown that the differential equation is determined by these postulates. The point of departure for another generalization is Schwarz’s s- function (cf. 2.7.2) which maps a half-plane on a triangle formed of three circular arcs. Koppenfels (1937, 1939) has studied the function which maps the area bounded by four circular arcs on a half-plane, assuming that two of the four angles of the curvilinear quadrilateral are 7/2 and the other two either 7/2 and 37/2 or 37/2 and 37/2. E. M, Wright (1935, 1940) has investigated the asymptotic behavior of the sum S Pla, +8,n) + (a, +80) 2” see eerate | nao V(p, +4.) °F, +H Qn) nt for large |2|. Here the B, and the 1 , are real, positive and 4 P 14+ 3 y- 6,50. 1 1 If all the », and 8, are equal to unity, this reduces to a multiple of hee ”' fieine’s hypergeometric series are discussed in section 4.8, For further generalizations see Chapter 5. The following conventions will be observed throughout this chapter: The values of the q parameters p,,... , Pa in Fa are always dif- ferent from 0, ~1, -2, «+++ Tf the variable z in oH ie is not equal to unity, we assume tacitly Jal0. Tf the variable z ina Fe is unity it will be omitted,184 SPECIAL FUNCTIONS 4.2 4.2, Differential equations The function oF is defined by 4,1 (1). The series was introduced by Clausen (1828) in the case p = 3,q = 2; the notation is that of Poch- hammer as modified by Barnes. If one of the a, is a non-positive integer, the series terminates; the cases in which one of the p , is a non-positive integer are excluded. Ifp=q+land (I) s = Rel, +e++ 0-0, 2a 4y)s then the series in 4,1(L) converges for all |z| = 1 if s > 1; it converges for all |z| = 1, z #1 if 12> s 20 and is divergent if s < 0. The proof is the same as the one given in 2L4L for , Fy. Let 5 denote the operator 2d/dz. Then u= ,F, satisfies (2) 186 +p,-N +++ (Ftp ,-N-~ 25 +a,) +++ (Sta) u=0, which is equivalent to the general equation of the type a (8) 3S 2™'(a,2-b,)D'v+a,v+z9DM vy =0 q2P ney or 4 (4) SB 2 "(a,z-b) Dvtav+z%(1-2D™vy=0 pagel n= where a,, ,, are constants, a, £0 and D = d/dz. For (3), 2 = 0 and z = oo are singular points, of which z = 0 is of regular type (cf, Poole 1936, Chapter 5); (4) is of Fuchsian type (cf, Poole 1936, p. 77) with regular singularities at 0, @, 1. For a set of linearly independent solu- tions for the neighborhood of z = 0 and z = © see section 5.4, There and in the greater part of the literature, only the " general” case is inves- tigated where none of the p , ard none of the differences p,—p,,a,~a, (r #5) is an integer. In the case p 1, In certain cases, quadratic and cubic transformations exist for ,F,, e-8., EF @, b, cr 2 Ms" tyya-blsane “0-37 F, a ae (Whipple 1927) and ® GF, [* 2b-a-1, 2-244; | l+a-b, l+an-c bab + 9/2 a a/3, a/3 + 1/3, a/3 + 2/3; ~27 z/[4(1-z)*] ee b,a-6+3/2 which have been proved by Bailey (1929). Bailey also gave a linear transformation of a nearly~poised ,F, of the first kind: Qa-La+%a-b-%y2 @ G-pe 44%, : 2 a a-%a+b+% ] 7 26-1, b+%,b~a-%s =(1-2)27) . =(1-2) Al b-% a+b 4% Burchnall (1949) has proved that a well-poised ,F, of argument x can be expressed as a sum of series ,F, of argument ~ (x ~ 1)7/(4), ‘There are also a few cases known where a ,,,F, can be evaluated for an argument # I, e.g., () OF a, 1+%a, b, c3-1 T(l+e-a)P(l+a~c) 43 | Ma, l+a-b, l+a~c TO +a)C0+a-b~c) (cf, Bailey 1935, p. 28, and also Bailey 1929). Some special cases: The formulas developed in section 4.4 and in the present section contain alarge number of results which cannot easily be proved directly and which have found much attention in the literature,45 GENERALIZED HYPERGEOMETRIC SERIES 191 For instance, 4+4(5) gives (with a=b =e =-n): LT ya\ |? _ cos Kam P04) Th + Yn) PC + 8n/2) 2" ® fen [0] ~~ [Pd +ka) Pasha al If-we substitute b = 1+a+7 in 4,4(8) and let x + 0 we obtain a, 1+ ac dye © sf Ee l+a-c, l+a-d, eee] Tt+a-c) TI +a-d TA +a-e)P(l+a-c “PO +aP+a-d-e)P+a-c-e)P+a A special case of this is the Dougall-Ramanujan formula 2,69), 62), G+s),+),a+2), () 142 2 _PG@+ DTG + DPG+DT@+ytz2+D —Ty+zt+DP@+z2+ DP @ty+D which is valid for Re(x + y + z + 1) > 0, Another consequence of (6) Gibaslesl 27-2 1)1 z-l_, &-De-9 3 ee 2 Cer E) ‘The series on the left-hand side converges if Re x > 1. A large number of other special cases is given by Bailey (1935, Examples pp. 96, 97) and Hardy (1923). The truncated hypergeometric series (@), 6) Gee). eee a ss soe, can be expressed in two different ways in terms of oF et Spe be| | =n a be and ae @), no z [* nn, l-n-c3 a | (emt l-n-a,1-n-5 If z = 1 we have yq(@ 6 ol = Ti+n+)DPG+n+) a, b,c+n; niT(a+b+ne a"2]catbtn+l]192 SPECIAL FUNCTIONS 45 _TQ+a-c) PA +b-e) “TQ=e)TG-c+a+d) xd p- nt Ong op [Poa l-b aes (@+DIC~D FAL 2cn+2 Tatn+ DEG +24) p [ore ene ta | eee ee alP@a+b-c+Dlernal 2 (cf. Bailey 1935, pp. 93, 94). Corresponding formulas exist for the trun- cated ,F, and a special truncated ,F, with unit argument (cf. Bailey 1935, pp. 94, 95). 4.6. Integrals The analogue of the Eulerian integral representations in section 2413, in the case of the general ,F, has been given by Pochhammer (1893b), Erdelyi (1937), Jef. also’ 5.2(2)] and by Pevnyi (1940) for integral relations of the type given in 24(2). The extension of Bames’ integral representation 2.1(15) to the case of a general ,F, cam be derived from the results in sections 5.3 and 5,6. Many definite integrals can be expressed by one or several series F, . In particular, the Laplace transform of a ,F,, p< 9, is 3 gly eae rast ato | MO a,3* | Py Pyreees Pg A formula by Shanker (1946) gives the Hankel-transform of a ,F,: Se PF Lp, Kp + m4 Ds Hlp—m 4D, Hlp+y4 Ds ye] cc+n+] sie x d,(2t) (2t)* de =(-1)" 2° x ,F, [p, 4p + m+); lp m+ Ds Kp tv + D5 -%27 1, where m=0,1,2,+4., and Re(p +y+1)>0, Re(p—m+1)>0, Re p> % For other special results see section 4.7 and Erdélyi (1938). 4.7, Various special results H. Bateman (1933, 1934) and Pastemack (1939) investigated the polynomials in z, Fz) = ,F, Cra+) 4+423 1; D n=0,1, 2, e085AT GENERALIZED HYPERGEOMETRIC SERIES 193 and Bateman (1936) studied the polynomials Z, (z) and J” (z) defined by Yol erate. Z,(2)=,F,emntl lite) a Tne leu) nlP@+DP@+14 hu)! Bateman’s results have been generalized by Pasternack (1937, 1939) and by S. 0. Rice (1939), The latter introduced HC, pv) = Fy (+n, n+ 1, 3 1, ps v) 278 Jte¥(2) = Fylonzut vt Le %us 2%), where n= 0, 1,2, «+ and &, p, », are complex variables but p #—n ~ 1, -n—-2,.... Rice found that H, Ag; p. ») Tp) i =< €-¥(1—2)P-E“* P (1 Qv0) di atte f L a) a wt) dt Rep>Re €>0, P,(z)=,F, ( 1, |z| <4 p40, +1, £2,--.. Krall and Frink (1949) have investigated a class of polynomials y, (a, 2). According to Rainville these can be written in the form ¥q(@ 2) = ,F,(-n, a +2- 1; -2) where n= 0, 1,2, «44, anda -140,-1,-2,..,. They, (a, 2) are orthogonal polynomials on the unit circle associated with the weight function (@-) ,F, Q;e- -2") They can be expressed in terms of Whittaker’s function W,,,(2) lef. 6,9(5)] in the form Iya 2) = OP 2H a, ny aye le Ds For recurrence relations, a differential equation, and other results for y, (a, 2) see Krall and Frink (1949). 4.8. Basic hypergeometric series The following account. of the theory of basic generalized hypergeo- metric series follows closely Chapter 8 of Bailey (1935). Let q be a parameter which in general shall be restricted to the domain |q| <1. We define Q) @),,,= (14) (Lag) (= ag*) «+ (l-ag™™") MHL eres (2) (@),.=1 for all a and g.[The notation [a], for (2), , (where the q is not shown explicitly) is even more common in the literature .] Then ae FJ Lada nden Dan on Goa. (3) ,&, Ade ie [ Pro see Py la Tete P dan is a function of z and of r +s +1 parameters ay, ...54,5Pys +++ 5 pid which reduces to yy yy very SZ Fe Parsee oy196 SPECIAL FUNCTIONS 4.8 if r=s +1 and q> 1. The function ,®, was first investigated by E. Heine (1878, p. 97-125). ,®, is called a basic hypergeometric series; it seems that the case r = s + 1 is the only one which has been dealt with. For basic hypergeometric series of two variables see section 5.14. The simplest case is that of ,,(a; z) which is a generalization of the binomial series, It may be shown that l-aq"z (4) ,®,(a; 2)= a n=0 and therefore (5) ©, (a; 2) ©, (6; az) = ,, (ab; 2) (cf. Bailey 1935, p. 66). Other elementary cases are: z ae cue (6) 9 291% 59 a= J age i te (DQ 0,(@-4)-95 z= 142 see err i — Kg, y 2" @ FTF Mw iam Ds a If we divide (8) by z¥ and replace q, z, by 92, g exp (2x) where x is real, the imaginary part of the series becomes So g?*4 sin(Qn+ Ix Kk Ce) en Ay nat ae PT (1-29 cos 2x +94) (1+ g@""")* (1-9)? Se ee ee (L=24?""" cos 2x +g?) where sn u denotes Jacobi’s elliptic function of modulus k, and where , K, and q are connected by (9) q=exp(-7 K'/K), (10) K=%a Fs, 451k), K' 4a (4, 4s 1k?) 7 \ € See oko: This result is part of the theory of elliptic theta-functions, see Chapter ll. Many of the results in the theory of generalized hypergeometric series have analogues in the theory of basic hypergeometric series. This holds in particular for the theorems on well-poised series. The analogue of Dougali’s theorem 4.4(8) is Jackson's theorewn.4.8 GENERALIZED HYPERGEOMETRIC SERIES 197 ap 6 a, qa", -ga", b,c, de, q-"3 9 o'7 Le%, -2%, aq/b, aq/e, aq/d, ag/e, ag" ** \ (0d) yy (a9/ed)yy (24/bd),y (04/be)py sce este (cq/6),,, (0970), y (aa/d), y (aq/bed), y where bede = a? g¥*! and N = 1, 2, 3, .... The effect of the presence of the four elements ga“, -qa”, a%, —a” in the function on the left is merely the insertion of the factor (1 - aq2")/(1 ~ a) in the general term of the series, Another important result (which is the analogue of a theorem due to Whipple) was proved by Watson (1929): a .o,| qa", -qa", c,d, e, f, g; a* g?/Medefg) see ertate’. 2°7| 2%, -a%, ag/e, ag/d, aq/e, a9/f, aq/g aq") (1 ~ ag"/fg) (1 — aq"/ge) (1 - ag"/ef) agq"/e) (1 - aq”/f) (1 — aq”/g) (1 ~ ag"/efg) aq/cd, €, f, 839 x ,o, : efg/a, aq/c, aq/d For more general results see W. N. Bailey (1935, 1936, 19474, b, 1948 a, b). A discussion of the state of the theory is also given by Bailey (1947a, 1948b). Among the consequences of (12) there are the basic analogues of the theorems of Saalschiitz 4,4(3), Gauss 2,1(14) and of 2,9(2) viz. of Pore) Day Can S'la, beq'""/d} My (d/o) y ® b, ¢3 d/be Pp (1 - dg"/b) (1 dg"/e) a d “LL Haq") A= dq" Joe) b; Sb; 2®, [*: ‘| = ,®, [ab/es 2] ,®, [“ “ or. A large number of identities can be derived from (12). Some of these are Euler’s identity re FS Cyr [gener ghimtn ye Fg”, n= net the Rogers-Ramanujan identities ea ie LS ance 1-4")SPECIAL FUNCTIONS 4.8 198 = Gag tt agit, no grintt) 1 +2 G-9d-q)--a-7 HR Q-g?t9-"— gt), n=0 and a result due to Gauss 1-g™ 14a F ghntetny, ( . nai? I l-g™ For the proofs and for other results see Bailey (1935). The basic analogue of Kummer’s formula 2.8 (47) has been proved by Daum (1942), The result is a, b;-g/b a ewe _2(aq/6) Uqa*) Vga") 2-4/6) * (ag) ORG) RGqa" 7b) A qa*/6) ye gntt ae-[ ie . n=o whereGENERALIZED HYPERGEOMETRIC SERIES 199 REFERENCES Bailey, W. N., 1928: Proc. London Math, Soc. (2), 28, 242-254. Bailey, W. N., 1929: Proc. London Math, Soc. (2), 29, 495-Sié- RE evra, Bailey, W. N., 1935: Generalized hypergeometric series (Cambridge Tracts, No. 32) Cambridge. Bailey, W. N., 1936! Quart. J. Math., Oxford Ser, 7, 105-115. Bailey, W. N., 1937: Quart, J. Math., Oxford Ser. 8, 113-114. Bailey, W. N., 1947a: Quart. J. Math., Oxford Ser. 18, 157-166. Bailey, W. N., 1947b: Proc, London Math, Soc. (2) 49, 421-435. Bailey, W. N., 1948a: Proc. London Math. Soc. (2) 50, 1-10. Bailey, W. N., 1948b: J. London Math, Soc. 22, 237-240. Barnes, E. W., 1906: Proc. London Math, Soc. (2) 5, 59-116. Bateman, Harry, 1933: Tohoku Math, J, 37, 23-38. Bateman, Harry, 1934: Ann. Math. 35, 767-775. Bateman, Harry, 1936: Duke Math, J. 2, 569-577. Bose, B. N., 1944: J. Indian Math. Soc, (N. S.) 8, 120-128. Burchnall, J. L., 1948: J. London Math, Soc. 23, 253-257. Burchnall, J. L., and T. W, Chaundy, 1948: Proc, London Math, Soc. (2), 50, 56-74. Chaundy, T. W., 1942: Quart, J. Math., Oxford Ser. 13, 159-171+ Chaundy, T. W., 1943: Quart, J. Math., Oxford Ser. 14, 55-78. Chaundy, T. W., 1943: Quarts J. Math., Oxford Ser. 14, 55-78 + Clausen, Thomas, 1828: J. Reine Angeu, Math., 3, 89-91. Darling, H. B. C., 1932: Proc. London Math, Soc. (2) 34, 323-339. Daum, J. A., 1942: Bull. Amer. Math, Soc. 48, 711-713. Dixon, A. C., 1903: Proc. London Math. Soe., (1) 35, 285-289. Dougall, John, 1907: Proc. Edinburgh Math. Soc. 25, 114-132. Erdélyi, Arthur, 1987: Quart. J. Math, Oxford Ser. 8, 267-277. Erdélyi, Arthur, 1938: Monatsh.Math, Phys. 46, 132-156. Erdélyi, Arthur, 1939: Monatsh.Math. Phys. 47, 87-103, Fasenmyer, Mary Celine, 1947: Bull, Amer. Math, Soc. 53, 806-812. Goursat, Edouard, 1888: Ann. Seis Ecole Norm. Sup. Ser. 2, 12, 261-285, 395-480. Goursat, Edouard, 1884: Acta Math. 5, 97-120. Hardy, G. H., 1923: Proc. Cambridge Philos. Soc. 21, 492-503. Heine, Emil, 1878: Handbuch der Kugelfunktionen, Theorie und Anwendungen, G. Riemer, Berlin.200 SPECIAL FUNCTIONS REFERENCES Hill, M. J. M., 1907: Proc. London Math. Soc. (2) 5, 385-341. Koppenfels, Werner von, 1937: Akad. Wiss. Wien, S. B. 146, 11-22. Koppenfels, Werner von, 1939: J. Reine Angew, Math. 181, 83-124, Krall, H, L., and Orrin Frink, 1949: Trans. Amer, Math, Soc. 65, 100-115. MacRobert, T. M., 1939: Philos. Mag. (7), 27, 579-581, and (7), 28, 488-492. Mitra, S. C., 1942: J. Indian Math. Soc. (N. S.) 6, 84-86, and 81-83; see also Math, Rev. 4, 141. Pasternack, S., 1937: Proc. Nat Acad, Sci. U. S. A. 23, 91-94, Pasternack, S., 1939: Philos. Mag. (7), 28, 209-226. Pevnyi, B. G., 1940: C. R. (Doklady) Acad. Sci. U.R.S.S. (N.S.) 28, 310-312. Pochhammer, Leo, 1870: J. Reine Angew. Math. 71, 316-352. Pochhammer, Leo, 1871: J. Reine Angew. Math, 73, 135-157. Pochhammer, Leo, 1888: Math, Ann, 32, 353-371. Pochhammer, Leo, 1888b: J. Reine Angew. Math. 102, 76-159. Pochhammer, Leo, 1891: Math. Ann. 38, 586-597. Pochhammer, Leo, 1893: Math, Ann. 41, 197-218. Pochhammer, Leo, 1893b: J. Reine Angew, Math. 112, 58-86. Pochhammer, Leo, 1895: Math, Ann, 46, 584-605. Pochhammer, Leo, 1898: Math, Ann, 50, 285-302. Preece, C. T., 1924: Proc. London Math. Soc. (2), 22, 370-380. Rainville, E. D., 1945: Bul Amer, Math. Soc. 51, 714-723. Rice, S. 0., 1939: Duke Math. J. 6, 108-119. Saalschiitz, L. 1891: Z. Math. u. Phys. 36, 278-295, 321-327. Shanker, Hari, 1946: J. London Math. Soc. 21, 194-198. Shanker, Hari, 1947: J. London Math. Soc. 22, 112-115. Smith, F. C., 1938: Bull. Amen Math, Soc. 44, 429-433, Smith, F. C., 1939: Bull, Amer Math, Sac. 45, 629-636. Thomae, J.,1869: Math, Ann. 2, 427-444. Thomae, J., 1879: J. Reine Angew. Math, 87, 222-349. Watson, G. N., 1924: Proc. London Math, Soc. (2), 22, XXXII- XXX] Watson, G. N., 1929: J. London Math Soc. 4, 4-9. Whipple, F. J. W., 1925: Proc. London Math, Soc. (2), 23, 104+114,GENERALIZED HYPERGEOMETRIC SERIES 201 REFERENCES, Whipple, F. J. W., 1927: Proc. London Math. Soc. (2), 26, 257-272. Whipple, F. J. W., 1935: Proc. London Math. Soc, (2), 40, 336-344, Whipple, F. J. W., 1937: Proc. London Math, Soc, (2), 42 410-421. Whittaker, E, T., and G. N. Watson, 1927: A course of modem analysis, fourth edi- tion, Cambridge. Wright, E. M., 1935: J. London Math, Soc. 10, 287-293, Wright, E. M., 1940: Proc. London Math. Soc. (2), 46, 389-408.CHAPTER V FURTHER GENERALIZATIONS OF THE HYPERGEOMETRIC FUNCTION 5.1, Various generalizations The classical hypergeometric series (fee 7 2 eevee ly 1. 2-y+D has inspired the investigation of many functions and series. In this chapter we shall be concerned with those generalizations which are commonly regarded as hypergeometric functions. Other generalizations, for instance, Mathieu and Lamé functions, will be found in other chapters. The generalized hypergeometric series (a,), GQ) hG@. 66) 8 ele nwo), where ie 3) @,= ee, (@),=1, (@,= aa+l+-(a+n-D, and where it is assumed that the parameters are such that at least one of the definitions (3) makes sense, has already been introduced in Chapter 4. Here we shall regard (2) as a formal power series and will pay no attention to questions of convergence. It is usual to define the most general (formal) hypergeometric series as a formal power series for which the ratio of two consecutive coef ficients is a (fixed) rational function of the index. More precisely, Go 8. os 2 nzo” cae Cael): where P and Q are polynomials; we assume that Qn+D=@4+NQ,Q+) where Q, is a polynomial such that P(n) and Q,(z + 1) have no common factors. The respective degrees of P and Q, are p and q. The factor 2025.1 GENERALIZATIONS OF THE HYPERGEOMETRIC FUNCTION —-208 ization of P and Q, in linear factors at once leads to the expression of (4) in terms of the symbol ,F,, 90 that essentially (2) is the most general (formal) hypergeometric series. ‘The differential equation satisfied by (4) can be written in terms of the differential operator (6) 5ex— wet. We have 6" = nx" and hence for any polynomial operator with constant coefficients f(6) x" = f(x) x". Thus (4) satisfies the differential equa- tion d (6) txP(8)~Q(S)}y=0 or 1P()~ 7-0, (hy =0, which is of order max(p, g + 1) and has singularities at 0 and © if p#q+l, and at 0, oc, and at a third, finite, point if p= +1. In connection with this generalized hypergeometric function there are two distinct problems: (i) to interpret (2) when p > q + 1 and the series is divergent for every x # 0, and (ii) to find fundamental systems of solutions of (6) for the neighborhood of every singularity. A further generalization introduces basic hypergeometric functions. For these see Chapter 4. Hypergeometric functions of two or more variables are similarly defined. For these see section 5.7. MACROBERT’S E-FUNCTION 5.2, Definition of the £ - function MacRobert’s E'-function arose from an attempt to give a meaning to the symbol ,F, when p> q+ 1. Forp lifp=q+1L For p> q + 1 we have (2) Elpsa,: 93 p,:x)= i Ta,-a) =) o——_ Pra) a =) Tt Pp,-a) a204 SPECIAL FUNCTIONS VF yaplays a, py ty ees ,- pth; Om Oy errr ar eeey ama, th CDP x] where |x| < Lif p= +1. The prime in I’ indicates the omission of the factar (a, — a,); the asterisk in F the omission of the parameter a,~ a, + 1; an empty product is to be interpreted as 1; and zero or negative in- teger values of the a, are tacitly excluded. Similar conventions will be observed throughout this chapter. The asymptotic expansion, for x + 2, - Ap - q+) aK ag x < 4(p - ¢ + 1) a, of (2) is given by the right- hand side of (1). Originally (MacRobert, 1937-1938),£ was defined by the multiple integral Ta, 3) Eps aj: 93 p,: x)= Gree a) 4 xu f, NO ad) Hdd, cae te promt x nA, O47) ‘aty \ “ty re Ady dy at vee oe A eas Bye 1 eater dd cL es [ * Ta) Otay Fa, ° where arg x| <1 p>q+ land the a, and p, are such that the integrals are convergent. The equivalence of (2) and (3) can be proved (Mac Robert 1937-1938). Other related functions are the P-function which arises when (2) is written as P (4) E(psa,:93p,:2)= x, Plas p-a,: 95 py: %) and two further functions, denoted by Q and H, which are multiples of P and E ae, (MacRobert 1937-1938). An alternative notation for E(p3 ay: q3 pi %) is E (ays vey Oy! Pys sees Pg 2 Several beecie cases of the E-function arise from (1) by means of the formulas developed or quoted in Chapter 4. From (2) and (3) some further interesting particular cases arise among which the most important are the expression of the modified Bessel function of the third kind, (5) (na) K(x) =e7* cos(vn) E(% +4 4 ~ v2 2x)5.2.2. GENERALIZATIONS OF THE HYPERGEOMETRIC FUNCTION 205 {MacRobert 1937-1938 (12)], and that of Whittaker’s W-function (6) TK-k-mTK~kam) xt eM, | (x) =EQ-k-m, 4-k+m: ix) [ MacRobert 1941(25)]. It may be noted that V, , (ix) W, , (ix) may also be expressed in tems of the E-function [ MacRohert 1941 (15')] as can other combinations. The £-function itself is a particular case of Meijer’s G-function [cf. 5.6(2))- 5.2.1. Recurrence relations A basic system of recurrence formulas has been given by MacRobert [ 1941, equations (20) to (24)]. The three most important formulas are CD) ay x Elay seers Gps pyr eves Pgt ®) =e Ela, ty ayy sees Opt Pps eve y Py tA) +Elay+Laptleesa,thpy thes pet hay d 8) Ely ees at Pry ene 9 Pgt AD =x? Ela, +l eee, a,th ppt lees pth) and (9) (py Vx E lay, e+ a, pyre + qi) = Ela,,...,0,:p Veer sees Ogi x) ee ee 5.2.2, Integrals A few integrals with hypergeometric functions may be evaluated in terms of the E-function. A typical example [ MacRobert 1937-1938 (21)] is ao) * eth) LF, (ay B; 8; -d) dr T() 277 r@ ) which may be proved by substituting oF, (a, Bs 8s ~X) e ae a So uPA (Ley? sar] dy E(a, B, y: 8: x) Rex>0, Rey>0, T'(6) '(8- p)206 SPECIAL FUNCTIONS 5.2.2 on the left-hand ‘side and using (3). The integral Sf e®® Fla, Bi ns OWE may be evaluated by complex integration; we integrate around a contour which is a rectangle with vertices at 0, 1, 1+ ik, ik(k > 0 )and which has indentations at 0 and 1 and then make the radii of the indentations tend to zero and k » «, using (10). [MacRobert 1937-1938 (23); there are also a few other integrals in this paper. ] Integrals with the E-function are discussed in Mac Robert 1941. The most important are the integrals of the Laplace or Euler type, but there are some more general integrals too. Typical samples are 1 tot) oa OF EE ays vee yt pay ee gt Ea) dE —mgagesa, MEG, y o00y Opt Pyy ves Pg 2 *) (12) JP eA AP ECG, cee yt Dyr eee Dgt A/A) GA Ely ose yyy Bt pps 000 9 Pgt®) Re B>0, (ot) (18) Jo e® Et Blayy vee 5 yt ys vee Pgs 3/E) dE =e¥"E(a + ys Bt Pps o0+ y Pg 28%) we OTE (ayy sees ys Bt Pas tease). If p=q+1, this holds provided that ¢= x inside the loop; ifp 1, the right-hand side reduces to 24 sin(Bm) Bays 01+ 4 5) Bi Py eee Pgi 2 (14) SP ATE LAN EL ayy soe yay Pyy eee Pg? +2] =P (o~B) Ela, 0+ 5 ayy Bt yy vee» Pgy 0 *) see errta! Re(B)>0, Re (a-f)>0. The proof of these formulas is based on the definition of the E- function; (1) or (2) and (3) are used according to the values of p and g. MEIJER’S G-FUNCTION 5.3. Definition of the G -function Meijer’s G-function provides an interpretation of the symbol ,F,5.3 GENERALIZATIONS OF THE HYPERGEOMETRIC FUNCTION 207 when p > q + 1; this interpretation is in complete agreement with the one given by MacRobert’s E-function. In addition, all significant particular solutions of a hypergeometric differential equation 5.1(6) may be ex- pressed in terms of the G-function. Originally (Meijer 1936), the G-function was defined in a manner resembling 5.2(2). Later (Meijer 1941c, 1946), this definition was re- placed by one in terms of Mellin-Bames type integrals (for which see section 1.19). The latter definition has the advantage that it allows a greater freedom in the relative values of p and g. Here we shall complete Meijer’s definition so as to include all values of p and q without placing any (non-trivial) restriction on m and n. We define o ese (fet) A, T;-s) A, T(1-a;+s) ads ma is i ra- te a, (Leas) j=att J where an empty product is het as 1,0 1 and either p lLaendeitherp > 9 orp=q and |x} >L208 SPECIAL FUNCTIONS 5.3 We shall always assume that the values of the parameters and of the variable x are such that at least one of the three definitions (2), (3), (4) makes sense. In cases when more than one of these definitions make sense, they lead to the same result so that there will be no ambiguity involved. The G-function is an analytic function of x; it is symmetric in the parameters @. 1 @,, likewise in @,4,1 +++) py in by, «++,» and inb, With (3) the idtegral can be evaluated as a sum of residues. If no two , j=1,+++>m, differby an integer, all poles are of the first order and = Ct A re,~s) Brat e-ap P i Pp _i ] oT se T'(a,-6,) x Fy, [1+ b,- 5 Lib, m dys eceg acces L4b, = 55 (DPA a] pPL y14b)-a,3 sy lta,-a,3 CTs In these expansions conventions corresponding to those of 5,2 (2) hold.5.3.1 GENERALIZATIONS OF THE HYPERGEOMETRIC FUNCTION 209 5.3.1. Simple identities If one of the a, j= 1, ... , 7, is equal to one of the b., j=m+1, visa q(orone of the b,, j= 1,...,m, equals one of the a, j=an+ 1, s+» P)y then the G-fimction reduces to one of lower order: p, g and n (and m) decrease by unity. mex, cet ee (eferrts) ™ prq>) is such areductionformula, and all others are similar. Chvious changesof the variable in the integral (1) give o cant |r \ gan re 2) cal: s) Om ( re). nm co To. (9) om =cm **) r (10) on(- 2 see evrote! = (Qn)Kethar ann gkp-katimayn 884 a, t,t tb, p24 4b, 4b, +% In (10) the duplication formula of the gamma function has been used, and there is a corresponding formula which uses the multiplication theorem 1,2(11) of the gamma function. The most important of these formulas is (9) because it enables as to transform a G-function with p > to one with p 1, @,\_1 Tibaey om ttn =m [97 | msq-l P a () (neers Th S-a4D- Mh (9-o)]y=0 Sax, jst a pat i dx Clearly, every differential equation of the form 5,2(6) may be reduced to this form by a change of variable. Equation (1) is of degree max (p, q) and on account of 5,3(9) we may assume p < q. The solutions of (1) have been investigated by Meijer (1946). If p p +2. Then, if is in the sector determined by (3) and (4), the p functions (5) Ge (xterra yr |The Fae rere Bpagt Fatty vee a, ) b b 1g hades es and the g — p functions (6) Gre (- Geomcn- ahd form a fundamental system of solutions of (1), If p = q, the point x = © is also a singularity of the regular type and, with the conditions (3), a fundamental system is given by (5). In this case x = (-1)?""~" is also a regular singularity, but no funda- mental system for the neighborhood of this point has been given in the literature. Ifp > q,5.3(9) may be used in order to reduce the differential equa- tion to the case already discussed; x = 0 and x = o interchange their a ") h=g, gt+h-.-,g+q-p+l role. In any case, it is clear from (1) that for fixed p, q ay) --. 5 Gp» Dyy sees bay all @ +1) (q+ D functions @. CoC hee] Osmsq Oxnsp satisfy the same’ differential equation. 5.4.1. Asymptotic expansions We shall assume p ¢ can be obtained from 5.3(9) by interchanging the role of x = 0 and x = o. Certain integer values of combinations of parameters are excluded, and so are a few212 SPECIAL FUNCTIONS 5.4.1 other exceptional cases, for example, those in which the coefficient of the dominant term of the asymptotic expansion is zero. The point x = 0 is a singular point of the differential equation (1), and the behavior of G (x)in the neighborhood of this point follows from 5.3(5)- We have (8) Grr (x)=O(k|4) as x90 \ where p 1 m+n>kp+%q, and larg x|<(m+n—%p—'q) am; or if (10) g=p+1, & is some integer, and largx—(m+n—p+2k—-Vn|<¥n As x > 0, p ¥p+khq n=0, jargx|<(m-%p-~q)m When +0 and p %p+%q and fargx|>(m+n-%p-\q) a or (ii) m+n ¢ %p+%q and no restriction on arg x. (13) Ifq =p +1, letk be an integer, and let Jarg x —(m-+n—p + 2k)n| < hm In this case we must have either (Gi) m+n2p+1 and k20 of kg p-m-n or (ii) m+n p and no restriction on the integer & When p =, the behavior of G(x) as x -» & follows from 5,3(6); it is that of a power of x. For more complete and more detailed results see Meijer 1946 $18.5.5.1 GENERALIZATIONS OF THE HYPERGEOMETRIC FUNCTION 213 5.5. Series and integrals Of the more involved functional relations for the G-function the mast important ones are series and integrals, Comparatively few series of G-functions have been investigated. The number of known integrals is, however, very considerable. Only a féw samples will be given here; for more details refer to the original papers, mostly by C. S. Meijer. 5.5.1. Series of G -functions A first group of series expansions in terms of G -functions consists of the four multiplication theorems (Meijer 1941 c), cer( ae [Or oe Pa Byseees by b - Gye @ poo bo ee vere, o Aan” Ce) A-1U1, Redr>4, al SL io =a" 3-0-4) ox( a,-l te ‘s 4) =a"? . +1) oo ] 2a a ' Re X>e%, 2 which express G(Ax) as an infinite series of G(x). Ifp £9 and m= 1" the restriction |A — 1| <1 may be omitted in (1) and a similar re- mark applies to (3) when n= 1, p>q. ‘The second set of important series consists of the so-called expan- sion formulas (Meijer 1946). These serve to express a G-function as a finite combination of G-functions with the same p, q, but changed m,n and are very useful in the investigation of the solutions of the differential equation 5.4(1). For instance, the fourth expansion formula (Meijer 1946, Theorem 5) expresses a G"" as a linear combination of v functions of the type G*, y functions of the type G*'", and another & ~~ v functions of the same type, under suitable restrictions on key Ly my My Ps Gy By Vo Breer Gry a, b +d we214 SPECIAL FUNCTIONS 5.5.2 5.5.2. Integrals with G -functions The most important integrals are those which express the properties of the G-function under integral transformations. The Euler transform is (5) e y1- Nee! Gan lin re 4 -Ta-p) opiate i ) and may be proved from 5,3(1) when (6) p+q<2(m+n), largx|<(m+n-%p—%q)n, (2) ReB-1 he=lewe,m Re (ata) ~1 ole q Better conditions have been given by Meijer (1936, equation (58)] in a particular case, 5.6. Particular cases of the G -function Clearly, a) Ee see eqrcta! 9 1 re) i=} Gay Glo = . f P Con 0,1-b,,-.., 1-8, Bre) Lob By ere a 0 re) pe . Gr = . uo(-2 7 0 : Bre and by comparison of 5.2(2) and 5.3(5) (2) Eps @,:93B,:%) (|i Bre 0B =GPl 1 +P216 SPECIAL FUNCTIONS 5.6 The importance of the G-function derives largely from the possibility of expressing by means of the G-symbol a great many of the special functions appearing in applied mathematics, so that each of the formulas developed for the G-function becomes a master or key formula from which a very large number of relations can be deduced for Bessel, Legendre, Whittaker functions, their combinations and other related functions. The material for the following list of special G-functions has been obtained mainly from several papers by C. S. Meijer. (3) G(x] a,b) = x4) 7 (ant) int () G2(e a, b)= 224K (ae) See errete, 6) 6% (- ba) wh eM Kha) (© GR ¢ |. ahibtert) oe WG) k=%(1+b+e)-a, m=hkb-%e % 4 0 2 (|, ) . wee ed (8) G2 é (9) GX | 4 26-4, 2b-a4+ han x? re atl) P(c-a+l xk tee, (x) 1 € . kea-%lb+c+), m=%b—%e see errata! ae Epa Cla bee year Chee a) (10) GR (x | a+%, a b, 2a—b) = 2-*[sin(a ~ b) w]7! 2-9? 2° (ap 282 2M) Ey g 2 2) — on" 2) X Jy gg) 2? 2) see ecrota! (11) GR (xl a ath, b b+ ha x¥ 46) Fae) (12) G2 (x | a, -a, 0, 4) = 2% (sin 2an)™" x [yg (ze™ 4) JS, (ze) — I, (ze) Tagg (z077*/4)) em 28? V45.6 GENERALIZATIONS OF THE HYPERGEOMETRIC FUNCTION 217 (13) G2 («| 0, % a sa) = 0% i7' (sin 2am)! xle@ttd,, (e771) J, (ze) — TET, (ze!) x Jaq (ze) pave (14) 6% (x| 8a -%, 4 -2-% a—%) = 20% (cos 2an)"! 29K K (2? 4) (T,(297 24) 4 I (2? 4] (15) 63 (x | 0, a-¥%, -a -% 4) 24% a K,, (27? 2") (J, (2°? 2) cos az - ¥,,(2°7 x) sin az] (16) G2 (x | -%, @- %, a -%, 0) = - 40% x K, (2%? x M4) (J, (2% x4) sin am + ¥, (29? x4) cos az] (17) CRG | a b+%, b 26a) = 04 24 PK, (2? 2) Si (gap 2 2) (18) GEG | aq a+%, b b+ N= Amat @FK, | (4x4) (19) C2 | a, a+%, b, 2a —b) = 29 a4 x? xKyg-42” V4 gritty Oe V4 en 7i/Ay % ; or. ete (21) Git ( “|, os, (x) T(x”) 0, a, -a (20) = a6 “|, 4, 4) = HOO A (22%) (23) G 2¢ : i “ y) eH A (24) 6% Oe Ya xt Jy ge) Yaga) “|b, a, 20-2)” bra bre 4 =a, (25) 6% ¢ ja, 0) = 2 21 sin an)" (I 2, (2) =o? (o¥]SPECIAL FUNCTIONS 5.6 (26) ao o( 7 |) = 208 T(x*) Ke) a . . y (27) aa: 4 ): n¥? (sin 2an)~" (I, (x#) - I? (x*)) a, —, 0 < . % (a+b) (28) G3 é es _) = Er i sO 22 a cos pean 29) G2 os See th nT Ay On (29) Clcens 1 (1,_,(2x") -~ (2x) a+h = Kt 2K wrote. | (30) 6% Gee Qn xt K2(x4) See & ; . a+% oe Pa ¥ wn oa cena .)" ae Male = ¥,,22°] (32) 6%, (x i = 221 -a =) (La +) S49 (22%) a, b, ~ ci (aay ct (2 2*% —) 2 a¥82-t [eos ale) 1" |, 2g -ba x 28 HO) (a) HO) (x) nw jerb- Tla+e)T(a+e,) see evrota! oe a) raed) xx! JFlata,at+q;a+b; +%, a (35) ox¢ i. Seay" rh xd, (8) J, (%) : (36) oz(: ce ) = 2n% xt Tye) Ky (0%) b, ¢, 2a -c, 2a— 0, ¥ ca a, b, ~b, 4) . x LH, (24) HD, =H eH, %)] a co (38) G2 ( ae ‘)- whe W224) F_, (22%) 0, 4 b, — (7) 6%5.6 GENERALIZATIONS OF THE HYPERGEOMETRIC FUNCTION 219 (9) G2 ee ee a 0, % ~ 20? 2 oe eee (30) 2¢ ja, b, — b, .) isin az sin br x [ebm (ok) HE) Get) — ebm (oH (nV) at oo ca | oeten = 2¢ ee) coe an cos bx xfer (ey HE, eH) +o MEH (eh) HO (%)] (42) G4 ( XW, (2ix®) Wy, (-2ix%) a,ath ) = 2 x9 Kye) Kye) Te Hee yee TE +b~ a T(%-3 0) Hb, “ ie 6, ee ee (43) Galt eee ee 6 ee) het x ,F,(a+5,,a4b,,a+b,, a+b; arc,,a4¢,,04¢,3—2) The following combinations of special functions are among those that may be expressed in terms of the G-function. (44) x Tle) = GH (hx? | he has 4 4) (45) x d(x) = 44 G20 (4-4 x8 ut aa vt Mutt, hum hv, +My Mv) (46) x" ¥,(x) = 246% («= hu-hv-’ Yu- hy, Yur hy Yp-ho-& (47) x# K(x) = "62 (Yx?| hp t hy, hp 42) (48) x#K (a) = at xGR (44 xt | Kot hp B+ hy + hy, — dv + Mu, W-Me+ Kw) 4 wr-v (49) e* K x) =n 6 (220 SPECIAL FUNCTIONS 5.6 % wmv, A+ hy + ep A+ hy + he Mp hv, 4 wt ev . Yate (52) H (x) - Y, (x) = 2°? cos vm ( “|, oo. ) (50) e* K(x) = 2° cos un G2 (> (1D) x# H(z) =24 6) ( 7 yp + hv + 6 7c tutte (« Yop + hy, A gto) (54) 2#[_(@) Ll =a! Meos uz GH (x= A+ hus py + bv + Yo, Mp Yr, op + av, 1 D4 ~ Yop — hv) (4 - Yop + Kv) 4+ Mp ot “oi (x a) : % 6) J2() =n" ol (: nav) % 0, v, - v, (55) S,, (2) = 2" 57) J,@) L,@=n* on (. (58) x7 J, (2) d,(2) aoe [x2 4+ hho, ho - [ Wutvto), ¥v+o-p, k(uto-v), el ‘ (59) ¥# T(x) d(x) = 1 242 ox(? Yueh hu Yos Mon was) 64 see ermia! 7 Kahn (60) x# J, (x) ¥,(x)=-2* GB (: ne _) (1) L,(2) K,(2) = 271 G2 (" % ¥, 0, =v5.6 GENERALIZATIONS OF THE HYPERGEOMETRIC FUNCTION 22] (62) x K, (x) d(x) = a % 262K 1 woa(Es (63) x? T(x) K, (a) = 271 x62 [2 x HO (x) HO (x) = 97? 2 cos ve Kaku on (42 s ¢ Yury Mum sa) 3 4+ Yn See ectate ! 2(2) 2 27 ak O®( x? : x! K2(x) ® oa(. oo ) Ker hr, Met Ys hoe a) Yo, Ko + % ] Kospto), Kr+o-p), Kuro»), Klo~r—p) (64 z & (66) 2° K, (2) K(x) = 271 of ko, hot x CM | a? - [ 4 (v+pto), 4(v+o-p) 4(u+o-v), ol (61) x#K (xe™**) K, (we 7#/4) = 248 gh L vee( (68) x'e“ W, (x) = 6B ( apy apr Ye nen ta-») L-k+1 m+l+%,l—-m+% 7. a Oe oe rue ne) kelel a -on(« ary) (70) e# Wy (x)= x Qk# -%k, %— Mk ea(ere|, Aten) ‘+m, ’- km, Yom, — 4m (2) e* W,, (2x) = xB gr Gt) ae 2 x os ( sce ecrata! T%+m-A TK ~m—k) K+ hh, 4+ %k ‘am, 4+ 4m, - 4m, %— Km,222 SPECIAL FUNCTIONS 5.6 ant TW@+m-DTG-m-b Wahlek %+¥%l-k ) (72) x", (2ix) Wy, (-2ix) = (73) 2" oR, x62 (x i" . = Me) x W (74) Fy, (a; b, ¢3 - x) = Toro 02(« AL 4+ Bl +m, Al =m, % Kl+k+1,%l-k+1 Wl+%, 4Wl+1 wl+m+%, wl—-m+% I-1,- ) Gee ectota’ a, -5 PETOrY “T@rOrard * oe ) see ecrota' (75), F(a, 6, ©, ds e, f, ly - 2) x ox(: Other similar formulas are contained in Meijer’s papers, in partic- ular formulas for Legendre functions and various generalized hyper geometric functions. -a,-b,-¢,-d HYPERGEOMETRIC FUNCTIONS OF SEVERAL VARIABLES 5.7. Hypergeometric series in two variables ‘The great success of the theory of hypergeometric series in one variable has stimulated the developement of a corresponding theory in two and more variables. Appell has defined, in 1880, four series, F, to Fy (cf. equations (6) to (9) infra] which are all analogous to Gauss’ F(a, B; y; x). Picard has pointed out that one of these series is intimately related to a function studied by Pochhammer in 1870, and Picard and Goursat also constructed a theory of Appell’s series which is analogous to Riemann’s theory of Gauss’ hypergeometric series. P. Humbert has studied confluent hypergeometric series in two vari- ables. An exposition of the results of the French school together with references to the original literature are to be found in the monograph by Appell and Kampé de Fériet (1926), which is the standard work on the subject. This work also contains an extensive bibliography of all relevant papers up to 1926; the list of references given in the present chapter is largely supplementary to Appell and Kampé de Fériet’s bibliography.5.7. GENERALIZATIONS OF THE HYPERGEOMETRIC FUNCTION 223 Hom (1889) gave the following general definition: the double power series a & Alatys anzo is a hypergeometric series if the two quotients A (2) —24u2 = f(m,n) and = gm, n) Aunts A are rational functions of m and n, Hom has investigated the conver- gence of hypergeometric series of two variables and established the systems of partial differential equations which they ‘satisfy. For the special hypergeometric series investigated by the authors already named, by Mellin, and by several others, A is a gamma pro- duct, that is to say it is of the form 8) y.,= WP (a, +ujm+v,n)/Pa) i where the a, are arbitrary (real or complex) constaats, and the u ; and v , are arbitrary integers which may he positive, negative, or zero. The question, then, arises, whether this type of series is the most general one compatible with Horn’s definition. Clearly, f and g must satisfy (4) f(m, n)g(m+ 1,n) =f(m, n+ 1) g(m, n) for all m, n = 0, 1, 2, ... , and hence identically in m andn, since each of the two sides is 4 nttot1/A qn3 and it is easy to see that any pair of rational functions of m, n satisfying (4) generates a hypergeo- metic series. Birkeland (1927) stated that every rational solution of the functional equation (4) can be decomposed into linear factors, and this would seem to lead to gamma products for A,,. Ore noted (1929) that the Birkeland theorem is not entirely general and gave (1930) a thorough analysis of the sational solutions of (4). From Ore’s result it can be deduced that the most general form of A, is of the form A= Rm, n) yy, a" where R is a fixed rational function of m and m, a and b are constants, and y,, is a gamma product, This is equivalent to saying that the most general hypergeometric series of two variables results from the opera- tion of a rational differential operator, R(x224 SPECIAL FUNCTIONS 5.7 on a hypergeometric series of the type E Ygq (ax)* (by). It would seem, therefore, sufficient to investigate series of the Horn- Birkeland type. 5.7.1, Horn’s list Horn puts F(m, n) G(m, n) (5) f(m, n) FG)’ (m,n) = Can where F, F', G, G' are polynomials in m, n of respective degrees Pp p's % 9'. F'is assumed to have a factor m+ 1, and G’ a factor n +1; F and F' have no common factor except, possibly, m + 1; and G and G' no common factor except possibly n + 1. The highest of the four numbers p, p', 9 9’, is the order of the hypergeometric series. Hom investigated in particular hypergeometric series of order two and found that, apart from certain series which are either expressible in terms of one variable or are products of two hypergeometric series, each in one variable, there are essentially 84 distinct convergent series of order two (Horn 1931, corrections in Bomgisser 1933). There are 14 complete series for which p = p’ = g=q/ = 2: , ain Pp (Bn a sn © FB wane) Gar ty", (2, 45 (By Bn oo yn (), (y'), mint M Fle BB ny en=), : (2 (0), By oa yan mint @ Fae’, BB yHN= > (yin Bete ga ye QD, 7), min! ( ( (B"),- (10) 6(@ BB's my) > yn Ban By n xty", mint! O FaByy's. sy= QD 6,la a’, B B's % ny, 0, (2")y Bana Bann ge yx", mint5.7.1 GENERALIZATIONS OF THE HYPERGEOMETRIC FUNCTION 225, mint (0, 4 Bs inn (13) Hy (a By 8% n=), mint a ae , (), 3 (4) H AaB Bom r= ) See Des “re (15) gla By y= Yo Data Pa sey, Dea, mint Cp, on (16) Ayla Br Bas = Sy ewan wt, Qn Oe QD H,@B ys y= (onan 7 (a), -, (y), (18) Hela By ey)= > a ty", (19) Hy (a By & % ¥) nen (By On ge on » (), min! and there are 20 confluent series which are limiting forms of the com- plete ones and for which p (@),, - =niB bee min (OA). H Glan =n Pla gm yn SS eae -) (3), mint ( (38) Hla 3% y= Clencn yt yt ©, mint 5.7.1 Iel r, ly| > s. We put max + = R, max s = S. In the absolute plane (7, s), the points representing associated radii of convergence lie on a curve C which is entirely in the rectangle 0 0. The application of this to the complete series of the second order gives the following table: Wu, v) Cartesian equation of C 27r? s* + lers t4(r—s)-1=0 ars =(s — 1)?228 SPECIAL FUNCTIONS 5.7.2 un) Cane (49) a rts # ad Qpt+y? Qty (50) a us r+(s- 9? ple + v) pty (Qnty)? Qe+y | 6D Se 4r=(s-1? KH v (2n+% (u+y-p) (52) =n | 14 16r?— 367s 287 ~s +2778") = 0 ale (53) a s?rts Qp-v (54) 4r=(s“t— 1)? Qp-v The domains of convergence of the various series can be represented as follows: ©,) ay ©» (0,0 oe Fy G, (0,0) Go) 0,0) G,0) ©, 0) G0) (0% (0, 0,» a,” 1, 3-22) (0,0) (4,0) (0, 0) , 0) (0, D (1,0) Regions of convergence5.8 GENERALIZATIONS OF THE HYPERGEOMETRIC FUNCTION 229 (1) 0) 0,1) (4, 4) (0,0) (%,0) (0,0) (%,0) (0,0) (4,0) ,1) ©, (4, 2y 2-2) hs 4) A, (0, 0) (4, 0) (0,0) (4,0) Regions of convergence In the case of confluent series either @ or ¥ vanishes identically, the region of convergence simplifies considerably, and any inequalities which may be necessary to secure convergence are recorded in (20) to (39). 5.8. Integral representations Basically, hypergeometric functions of two variables, as the corre- sponding functions of one variable, can be represented either by the Euler-Laplace type or by the Mellin-Bames type of definite integrals, The latter type invariably, and the former type with an elementary inte~ grand mostly, leads to double integrals. Since double integrals are some what untractable and not very wel] suited for the integration of differen~ tial equations, it is natural to seek for single integrals representing the functions. Such representations can be found in every case, but the integrand in most cases contains a hypergeometric function of one vari- able, or even a product of such functions. The large number of functions in Hom’s list makes it impossible to give a complete list of integral representations here; integral re~ presentations will be given for Appell’s functions only, but it should be noted that all of Horn’s functions have similar representations and that230 SPECIAL FUNCTIONS 5.8 many of these have been given in the papers quoted in the list of refer ences, Integral representations are useful in connection with the analytic continuation of hypergeometric series in two variables, their trans- formation theory, and also for the integrationof hypergeometric systems of partial differential equations. 5.8.1, Double integrals of Euler’s type The integral representations CG) Tiare )Ty-B-B') x Sf wht "8 (w=) 77F8 O81 = ue = vy)" du do nc ReB>0, ReB'>0, Re(y-8-8B')>0, LOG") TATBITG-BPG -B9 SESE Wht vB = a8 = 0978" emo) dado Re B>0, ReB'>0, Re(y-f)>0, Rely’ -B')>0, Q) F(a, B Bays Y= (2) F, (a, Bs B', yy y's % Y= and Ty) TPP(B')TG-B-8') . Ek Beet Oeet a = paternal ao, v0) atest x (L-ux)"@(1 = oy)" du dv Re B>0, Ref'>0, Re(y-B-B')>0 are easily obtained from the series by using either Euler's integral of the first kind for the beta-function, or the corresponding double integral (Appell and Kampe’de Fériet 1926, Chapter II). The function F, is much more difficult to handle and does not seem to possess a very simple integral representation. Of the various double integrals proposed, perhaps the simplest is Burchnall and Chaundy’s (1940, equation (68)] @) Fy(aa’,B, B's m= rOry’) T@T@rG-alG’ A) x Lf ut oH L784 (Lao) At 5 (4) F(a, B, y, y'3 x19), y(1-x)] =5.8.2 GENERALIZATIONS OF THE HYPERGEOMETRIC FUNCTION 231 x (La a) 907 HL a py BOVH9 "#9 ae = vy) 97 PN dado Re (a)>0, Re(B)>0, Rely-a)>0, Re(y'-£)>0. In all these integral representations it is assumed that |x| and {y{ are small enough to make both series and integrals converge. 5.8.2. Single integrals of Euler's type Picard has pointed out that F, can be represented by a single in- tegral in the form (5) F(a, B,B', ¥3% ¥) TO) tee 2 =! ees ee ate ee ee B @re=a dee (l-u) (1-ux)F (1-uy) du Rea>0, Re(y-a)>0, This representation has the great advantage that it can easily be con- verted into a contour integral valid for negative Re a and Re (y ~ a) and that it is the best tool for the complete integration of the system of partial differential equations connected with F,. Equation (5) sets up a factorization, into a function of x and a function of y, of F, by means of Euler’s transformation, and the great usefulness of (5) is due to this property, ‘There are corresponding relations for F, and F, (Erdélyi, 1948) T@)P(e')TA-@ (2.72)? xf C0° (D7 FU, Bs ys2/t) Flp’, B's y's y/o] dt where p +p’ = a+ 1, and the contour of integration is a Pochhammer double loop (1 +, 0 +, 1-, 0 -) such that |t| > |x| and [1-2] > |y| along ity and (2) F,(a, a", By B' ys x, y) = (2ni)* DL - p) P(1- p') TG) fee ene eal (Fe 1) eat x F(a, B; p; tx) Fla’, B's p's 1-2) yl de where p +p! = y In either case a special choice of p reduces one of the hypergeometric functions in the integrand to an elementary function, but with general values of the parameters it is impossible to reduce the whole integrand to elementary functions. The function F, proves again more difficult to-handle than its fellows: no single integral is known which factorizes F,. On the other hand, the ©) F,4BB' yy se N=232 SPECIAL FUNCTIONS 5.8.2 integral representation [Erdélyi 1941, equation (3)] (8) Fy(a, Boys v's % y) = (208) *# PO) Py) P@-y~y') xf WH Ot tH .0°} yoy Gg nny’ x Fla Byty' +1; %/t+y¥/1-9) dt, in which |x/e + y/(2 ~ #)| <1 along the contour, is reasonably simple and useful for the integration of the system of partial differential equa- tions associated with F ,, 5.8.3. Mellin- Barnes type double integrals Following Appell and Kampé de Fériet, the four integral representa- tions can be summed up as Ty) D(a) P°(B) (221)? x LE LEM OPE) PCO a)! yds at (9) Ox, y)= where the contours of integration are indented in the usual manner (cf, section 2,1.3). © and W are given in the four cases as B(x, y) Ws, t) Tla+s+i)T(B+s)T(B' +2) TB )TG+s+0) Tla+s+oT(B+s) PB’ +) Ty’) rp) My+s) Ty! +) Tla+s) P(e’ +) T(B+s)T(6' +0 Ta Mp )rG+s+0 Mats+)T(B+s+e)T(y') Tits) PG +0 (10) F,(a, B, B's ys % ¥) (11) F(a, B, B', ys ¥'s 5) (12) F(a, a’, B, B', vs % y) (13) F, (a, Bs ¥» ¥'3 % ¥) Integrals of this type were used in Mellin’s investigation of the hyper geometric functions. 5.9. Systems of partial differential equations The series A, x" y", where A. 4, ,/A,,= Flm m)/F'im, n),59 GENERALIZATIONS OF THE HYPERGEOMETRIC FUNCTION 233 Ants /A an = Gim, n)/G'(m, x) and F, F', G, G', are polynomials as in 5.7(5), satisfies a system of linear partial differential equations which can be written in terms of the differential operators a oy a (1) 6=%— and 8'= ax as (2) [F'(6, 8!) x71 - F(8, 8] z= 0, 1G! (8, 8')y7* - G8, 8'Nz=0. In what follows we shall restrict ourselves to hypergeometric functions of the second order, in which case we find two partial differential equa~ tions of the second order. The two equations are certainly compatible (since the hypergeo- metric series satisfies both of them), and from the general theory of such systems (cf. for instance , Appell and Kampé de Fériet 1926, Chapter IM) it follows that they have at most four, and possibly less, linearly independent solutions in common. A closer investigation shows that the systems of partial differential equations associated with the eight series 8) Fy, Gy, Gq, ®,,0,,0,, 1, and DP, of Hom’s list have only three linearly independent solutions, while all the other 26 systems have four independent solutions each. However, in the case of the systems associated with (4) Gy, Hy, Hes Hey By one of the solutions is a comparatively trivial elementary function of the fom 2? ¥°, with the following values of p and g: Series Pp o | 1 fe | 6, ~ Flat 20") = Peete’) 6) | Hy, By | a-2y42 @| A,B, ap ~a-3e, | In the following list of partial differential equations z is the unknowz function of x and y, Oo az ate ae Se nies oy a ay? (9) x(1-x) r+y(1- x) s +[y-(a+B + 1) x) p— Byg - az =0 yQ-yt+x(1-y) s+ly-(a+B'+ Dylg-B'xp - oBz =0f234 SPECIAL FUNCTIONS 5.9 (10) x(1-2) r-ays +ly~ (a+ B+) x] p - Byq~ aBz =0 : y(L~y) t-xys +ly'= (a+ B'+ Dy] g- B'xp - oB'z =0 ee (1) xQ-2)r+ys 4ly-(a+ p+ VD 2]p az =0 he y(l-y)t4as + ly (a's B'+ Dylg-a’p’z=0f (12) x(1-2)r—y* t-2ayst[y-(a+B+Dalp -(a+B +1) yq- az =0 yQ-y)t—x?r-2Qxys +[y'~(a+h+Dy]q —(a+f +1) xp - aBz=0 (13) x(L+2) r—ys —y?t+[1-B+(a+f'+ Dzlp +(B'=a-lyq+ oB'z =0 yy+Yeons—2? r+ (1-8! +(a+Br+Dylg +(B-a—1) »p + aBz=0 (14) x(L+2)r—y(1+2) s4+[1-B4(a+p'+D x]p ~ayq + oB'z=0 C y(lt+y)t~n(1+y) s+[1-B'+(a'+B+Dylq] 2 ~a'xp+a'Bz=0 ee (15) (14 42)r-(42 + 2D ys+y?2t+(l-a+(4a'+6) a] p -2a'yq+a'(a'+ I) z=0 y(L+ 4y) t—x(4y+ 2) sx? r4 [1-04 (4046) ¥]19q -2axp +ala+ 1) z=0 (16) casa at As ba Dale to Bo Dag ate Oh ~y(Lt+y)t+x(l-y) s+[a-1-(B+y+l) ylq+yap- Byz =0 . Cy, (17) x(2-Dr -xys+lle+ B+ eee ba yly +L)e xs +[L-at (y+ 5+ Dy] q+ ydz =0 a5 (18) x(1- 4x) r+y(1- 42) s-y? t+ [y-(a+ 6) x] p - AXa+Dyqg-aa+z=0 y(l=-y)et+x(1- 2y) s+ ly-(a+h+Dylq — 2Bxp - aBz =0 i.5.9 GENERALIZATIONS OF THE HYPERCEOMETRIC FUNCTION 235 (19) x= 4x) r= days -y?2 0+ ly-(4at Da p - Gat Dyg-aa+)z=0 7 yQ-y)t-2xys +[6-(a+ 8) y] g- 2Pxp -apz-0) “4” (20) x(1+ 4a) r-y(1- 42) sty? e+ fl -y+4le+ Dal py +(8a+2yq-ala+])z=0 y(l-y) t-xys + 2x? r? +[y-(a+B+Dylq Mee +(2+a~ 26) xp - afz =0 (Q1) «(1+ 4x)r-(1+4x) ys +y? t4+[1- B+ (4046) x] p’ -2, Dz=0 ayg+ ala+ 1) z 7 y(l+y)t-x(2+y)s+[l-at+(Bry+Dylg — yxp + Byz =0 (22) x(1— 4x) r+ days ~y?2+[8-(4a+6) 2] p + 2ayq-ala+1)z=0 y(t+y)t- 8xys +{t-a+@+y+D ylq - yap + Byz = 0. (23) ee ee) yttxs +(y-y)q-xp-az=0 a ys (24) xrtys +(y-2) p-Bz=0 yttus+(y-y)q-Biz=0f ? (25) ar +ys +(y-2)p~Be=0 |) yl+ustyq~z=90 e (26) x(1—x)r~xys +[y-(a+ B+ 1) 2] p— Byg- aBz =0 7 yt+(y'~y)qg-xp-az=0 1. (27) + y-) p-yq-az=0 yt+(y'-y) g-xp-az=0 2? (28) x(1-x)r+ys +ly-(@+ B+ 0 x] p - aBz=0 z yttxs+ly~y)g-a'z=0 " (29) x(1-2) r+ ys +ly-(a+ B+) 0, x+y <1 with the weight function x71 y¥ ~!, The series @) x" y"FiC¥m, -4n, 4-4, kn, —m-n—hs +h; 27%, ¥7*) (which can be expressed also in terms of F, ) and (5) x" y" Fyl-Km, Yn, 4m, 4—-n, hs +h; @? +y?-D 27%, G+y?-Dy?] occur in the investigation of hyperspherical harmonics. All these special cases are investigated in the monograph by Appell and Kampé de Fériet. 5.14. Further series Hypergeometric series of higher order than two, in two variables, have been studied by Mellin, Birkeland, Kampé de Fériet (cf. Appell and Kampé de Fériet 1926 Chapter IX), and by Burchnall and Chaundy (1941), Burehnall (1942), and Chaundy (1942), Hypergeometric series in three variables have been investigated by Horn (1889); series in n variables by Lauricella (Appell and Kampe’ de Fériet Chapter VII) and by Erdélyi (1937, 1939a). Particular cases of Lauricella’s series occur in the investigation of hyperspherical harmonics. An extension of basic hypergeometric series to two variables has been given by Jackson (1942, 1944).246 SPECIAL FUNCTIONS REFERENCES Appell, Paul and M, J. Kampé-de Fériet, 1926: Fonctions hypergeémétriques et hypersphériques. Polynomes d’ Hermite. Gauthier-Villars. Bailey, W. N., 1935+ Generalized hypergeometric series (Cambridge Tracts in Mathematics and Mathematical Physics No. 32) Cambridge, Barnes, E. W., 1907: Proc. London Math, Soc. (2) 5, 59-116. Barnes, E. W., 1908: Proc. London Math, Soc. (2) 6, 141-177. Bames, E. W., 1908: Trans, Cambridge Philos. Soc. 20, 253-279. Birkeland, Richard, 1927: C. R. Acad. Sci. Paris 185, 923-925, Borngiisser, Ludwig, 1933: Uber hypergeometrische Funktionen zweier Ver- ahderlichen, Dissertation, Darmstadt. Burchnail, J. L., 1939: Quart. J. Math, Oxford Ser. 10, 145-150. Burchnall, J. L., 1942: Quart. J. Math., Oxford Ser. 13, 90-106. Burchnall, J. L., and T. W. Chaundy, 1940: Quart, J. Math, Oxford Ser. 11, 249-270. Burchnall, J. L., and T. W. Chaundy, 1941: Quart, f. Math, Oxford Ser. 12, 112-128, Chaundy, T. W., 1942: Quart. J, Math., Oxford Ser. 13, 159-171, Erdélyi, Arthur, 1937: Akad. Wiss. Wien, S-B Ila, 146, 431-467, Erdélyi, Arthur, 1939: Proc. Roy. Soc. Edinburgh, 59, 224-241, Erdélyi, Arthur, 1939.a: Nieuw Arch, Wiskunde (2) 20, 1-34. Erdélyi, Arthur, 1940; Proc, Roy, Soc. Edinburgh 60, 344-361. Erdélyi, Arthur, 1942; Quart. J. Math, Oxford Ser. 11, 68-77. Erdélyi, Arthur, 1948: Proc. Roy. Soc. Edinburgh Sect. A 62, 378-385. Horn, Jakob, 1889: Math, Ann. 34, 544-600. Horn, Jakob, 1931: Math, Ann 105, 381-407. Horn, Jakob, 1935: Math, Ann, 111, 638-677. Hom, Jakob, 1936: Math, Anum 213, 242-291, Hom, Jakob, 1938: Math, Ann. 115, 435-455, Hom, Jakob, 1938 a: Monatsh, Math. Phys. 47, 186-194, Hom, Jakob, 1939: Monatsh. Math, Phys. 47, 359-379. Ince., E, L., 1942: Proc. Roy. Soc. Edinburgh, A 61, 195-209. Jackson, F, H. 1942: Quart J. Math, 13, 69-82, Jackson, F. H., 1944: Quart. J. Math, 15, 49-61. MacRobert, T. M., 1937-38: Proc. Roy. Soc. Edinburgh 58, 1-13,GENERALIZATIONS OF THE HYPERGEOMETRIC FUNCTION 247 REFERENSES MacRobert, T. M., 1941: Philos, Mag. (7), 31, 254-260. MacRobert, T. M., 1942: Quart. J. Math, Oxford Ser. 12, 65-68. MacRobert, T. M., 1948: Philos Mag. (7) 39, 466-471. Meijer, C. S., 1936: Nieuw Arch, Wiskunde (2) 18, 10-39. Meijer, C. S., 1940: Nederl, Akad, Wetensch., Proc. 43, 198-210 and 366-378. Meijer, C. S., 19414: ibidem 44, 82-92, 186-194, 298-307, 442-451 and 590-598, Meijer, C. S., 1941b: ibidem 44, 435-441 and 599-605. Meijer, C. S., 1941¢: ibidem 44, 1062-1070, Meijer, C. S., 1946: ibidem 49, 344-456, 457-469, 632-641, 765-772, 936-943,, 1063-1072, 1165-1175. Ore, Oystein, 1929: Cr R. Acad, Sci, Paris, 189, 1238-1240. Ore, Oystein, 1930: J. Math, Pura and Appl. (9) 9, 311-326.CHAPTER VI CONFLUENT HYPERGEOMETRIC FUNCTIONS 6.1, Orientation If we put z= x/d in Gauss” hypergeometric series For jo 2 ee oe 1-2-cle+ in which we assume that neither a nor ¢ is zero or a negative integer we obtain a power series in x whose radius of convergence is |b| and which defines an analytic function with singularities at x = 0, b, and co. As 6 + o, the limiting case will define an entire function whose singularity at x =» is a confluence of two singularities of F (a, 6; 0;x/b) In this manner we are lead to Kummer’s series In the notation of generalized hypergeometric series 4,1(1) this is 1 F(a; ¢; x), but in this chapter and in the following two chapters (1) will be denoted by Humbert’s symbol Bla, o; x). Sometimes the notation f(a, c, x) is also used. ‘The series (L) satisfies the differential equation d*y dy qrte-a fp -an0. The substitution (82) yar et®z, a=%—-x+p, cH lt+2Qp (2) « reduces (2) to Whittaker’s standard form dz ee (4) He 6Cx +Tt =0. Either of the two equations will be called a confluent hypergeometric equation, and any solution of either of them a confluent hypergeometric function; a and c (or « and y) will be called the parameters, x the variable. 2486.2 CONFLUENT HYPERGEOMETRIC FUNCTIONS 249 hh this chapter, the investigation of confluent hypergeometric func- tions will be based on (2), but some of the more significant results will also be given in Whittaker’s notation. 6.2. Differential equations Equation 6.1(2) is a homogeneous linear differential equation of the second order whose coefficients are linear functions of the indepen- dent variable; and it can be shown that every such differential equation can be integrated in terms of confluent hypergeometric functions. Let Che d; WM (@,x+5) = +(jx46) 2+ @r+bdy=0 be the differential equation. Ifa, =a, = a, = 0, the equation has con- stant coefficients and can be integrated by elementary functions. This case wil] be excluded. The transformation Q) yse™z, x=dEtp, (AF 0) carries (1) into d*z @) (06+ BQ) yee +(e £4 By) dz a (a, €+ B,) 2=9, where (4) ay=ag/A, a,=A'lhh a,=dA(h), Bo=(a.utb)/M, By= (AA) +B (V/A, B, = pA(h) + Bh), AQ) =a,h?+a,h+a,, A'(d)=2a,h+a,, Blh)=b kh? +b,ht+b,, B'(h)=2bh+d,. If we can determine A, 1, and h, so that a,p+b,=0, a,+AA'(A)=0, A(h)=0, then (3) reduces to 6,1(2), In other cases a change of the independent variable will reduce (3) to a confluent hypergeometric equation or to the Bessel equation. The results are set out inthe table on the follow- ing page, where 5(a, c, x) stands for any solution of 6,1(2), and C, (x) for any solution of the Bessel equation ady wy (5) x on ges (r-v)y=0. Ix x6.2 SPECIAL FUNCTIONS 250 € 9% CV?) == HY an 1D yh Tg] Oe eee |e yl) 9] =o S ong (499 D09 PE | 0% to ox | “Cnay/y) ,F-K =D (qa)/o~= 4 : GP1K ME I =. ox'o o=% | (2) At : (a ® Cozy) a =? ce otto : 222 (Cq'o = '9°n) = ' Elon oF 0 ¢*o°op —to= a ee lee | @ VA a=> ‘ox? suoyouwseg z ET x y suordunssy — z 1 one pe z, ‘ °. « ‘ ep qtr vasa 7 Mv So+y'oe va(yy W+RY=¥ ‘ey 2=h =k Rg 4 x2) + L('q 4 x!o) + ,£(Cq +n) Jo uoronpay6.2 CONFLUENT HYPERGEOMETRIC FUNCTIONS 251 The reduction of (1) to the confluent hypergeometric equation is not unique, for 6,1(2) can be transformed into an equation of the same form in several ways. If we put (10) x=Aé, y=xPeMz, the differential equation 6,1 (2) tranforms into d?n dn age tle# 2p (= 2h) a8) +[plpte-D/E-Ma~he + p-2hp) +A? Ath 1) El =0, and this is the differential equation of S(a, y, €) if (12) p=0 or l-c, &=0 orl, d(1-2k)=1, =Ma-he)+p, y=e+2p ql) é Thus, counting the identity transformation, 6.1(2) can be transformed in fow different ways into an equation of the same form. The confluent hypergeometric equation is a homogeneous linear differential equation of the second order with a singular point of regular type at x = 0, and with coefficients which are regular for all x 4 0 (includiag x = 00). Every such equation can be shown (Tricomi 1948) to be of the form a’y bY dy Boy 2 (ord) 2s (orb ed) g00, and it is easy to see that [apart from the trivial case when (13) can be integrated in terms of elementary functions] the integration of (13) leads to confluent hypergeometric functions or to Bessel functions, We have to distinguish between two cases of (13). Ifa? £ 4a, we have (14) y a278/? 8/2 wlk, py E), where (15) x =(B-%ab) (a?- 4a), p= %[(b- 1)? -4y]%, &=(a?~4a)* x, and w(x, 2, x) is any solution of 6.1(4); and if a? = 4a, we have (16) y = 2¥7P/2 6/7 C (8), (13) where (17) v=((6-D?=4y]*, &=20(8 - Kab) x]*. The equation d* dy (is) X2-22 4 (axes BY XX + (0X4 GHP +K)y =0. xX? ax252 SPECIAL FUNCTIONS 6.2 in which p # 0 is any number, is reduced to (13) by the substitution (19) =X", a-A/y b=(B+v- DA, a=Dv?, B=Gv%, yok and hence can be reduced to the confluent hypergeometric equation. A more general form can be obtained from (18) by patting (20) r-¥e0[ feo] Y satisfies the equation 2 (on x# 22, (axon By 2g x4 [xr scxres dx? dX ; d¢ eee Bs 24X— =0, +(AXP+B- Dh 4 G24 “ty 0 For instance, with ¢(X) = 4X7 it follows that d*z dz ax? + (AX? + BX7+ Oey + (DX + EXP +4 FX™ + GXP4HX°+K) z= 0 can be integrated in terms of confluent hypergeometric functions pro- vided that (23) E=%AB, F=\%B?, H=¥%B(C+0-D. At this point it is worth mentioning that Bessel functions themselves are special confluent hypergeometric functions (cf. section 6.9.1) so that in all non-trivial cases the differential equations of this section can be integrated in terms of confluent hypergeometric functions. ‘The principal linear differential equations of the second order which can be integrated in terms of confluent hypergeometric functions are (1), (13), (18), and (22), the last under conditions (23). (22) x? 6.3. The general solution of the confluent equation wear the origin We shall investigate the confluent hypergeometric equation in the form d® d) Weed 2 -ay=0. One solution of this equation is Q) y,= Og, ¢; x),64 CONFLUENT HYPERGEOMETRIC FUNCTIONS 253 and the transformations 6.2(10), 6.2(12) give three further solutions, viz, (3) y, 221° @(a-c 41, 2-c; x), (4) y,=e* B(e-a c5~x), 6) y,=2'e*O(1-a, 2-0; -2), From their behavior at the origin it follows that y, and y, are line- arly independent if ¢ is not an integer so that (in this case) the general solution of (1) may be written as (6) y=dy, + By, where A and B are arbitrary constants. The exceptional case of an in- teger c will be discussed later (cf. section 6.7.1). On the other hand, both y, and y, are solutions of (1), both are regular at the origin, and have the value unity there. Since (with a non= integer c) the differential equation cannot have more than one such solution, we must have y, =7, or (7) ®(a, ¢; x) = e* O(c —a, c; =x). This is known as Kummer’s eransformation. Similarly y, = y, by the Kummer transformation, Kummer’s transformation is a limiting case of Euler’s transformation F (a, bs ¢3 2) = (1-2) Fle ~ a, 6; ¢; zz - DI of the hypergeometric series (cf. 2,1(22)], and is a very important relation. Our proof assumes that ¢ is not an integer, but the trans- formation holds, by continuity, for positive integer c. ify, andy, are any two solutions of (1), their Wronskian, d ge cae must be of the form K__ e* x ~*, where the constant K can be evaluat- ed by using the first two terms of the series expansions of the y’s. We find (9) Wg = Way = yg =~ Wyq = (1-0) x8 e% All the other Wronskians of the four solutions vanish identically, 6.4, Elementary relations for the ® function As in the theory of Gauss’ hypergeometsic series, the four functions @ G@)=@+Le;x), Oa-)=O(-1,¢;%), O(c) =O(a,c+1;x), O(c -)= O(a, c-1;x)254 SPECIAL FUNCTIONS 6.4 are said to be contiguous to ® = ®(a, c; x), The function ® andany two functions contiguous to it are linearly connected. The six formulas describing these connections can be derived from Gauss’ relations between the contiguous functions (cf. section 2,1.2) and can also be verified by comparing coefficients of like powers of x. The formulas are as follows: (2) (¢- a) O(@-) + (2a~c +x) 6~a G(a4)=0, (3) ele -Y) O(c -)- cle -1+x)O+ (c ~ a) x O(c +) =0, (4) (a-c+)) @-a (a+) +(c - 1) O(c -) =0, (5) ¢®-c @(a-)-x O(c +)=0, (6) c(a+x) O-(c—a) x @(c +) ~ae G{a +) = 0, (7) (a- 142) 6+ (c ~a) O(a-)-(c-) O(c -) = 0. These relations are not all independent. From two suitably chosen ones, e.g., from (2) and (4), all the others follow by simple operations. ‘Any function O(a +m, c +n3x),m, n, integers, is said to be asso- ciated with (a, c; x). By repeated application of the relations between contiguous functions, it is easy to prove that any three associated func- tions are connected by a homogeneous linear relation whose coefficients are polynomials in x. By means of term-by-term differentiation, we have d a 8) —O(, ¢5 )=— O@+lc+h 2) dx e or @' = (a/c) O(a 4, c +), By repeated application, the derivative of any order of © is an associated function, and hence any three derivatives are connected by a homogeneous linear relation whose coefficients are polynomials in x, The differential equation 6,1(2) is the simplest exe ample of such a relation. Combining (8) with the relations between con- tiguous functions, we have [© - ®(c-1)] (9) oa (oa) -e1-(E- ) O(c N4O= : c Further useful formulas are: ap dx” (10) (a), O(a, c;x)=—* B(a+n, c +n; x) ©, n (ul) = [x2 O(a, oj I = (a), 27? O(a tn, o; x) fe65 CONFLUENT HYPERGEOMETRIC FUNCTIONS 255 a 02) [xe7! &(q, ¢; x)] =(-I)"(1—e), x67!" O(a, c —n; x) (c-a), a3) = [e~* ®(a, ¢; x)] =(-1)" e7* O(a, c +n; x) i) = [en# 2-21 Ba, 05 MI =(e =a), em xe! G(a=a, 65 2). 7" Tla+n) Tla+n (= (= ala+D ++ (a4n- 1) = n=1, 2,3, 0008 Basic integral representations It is known that homogeneous linear differential equations whose coefficients are linear functions of the independent variable can be in- tegrated by Laplace integrals. The integral representation The) ph yet yyene! dy (1) ®(@, ¢; x) = ————___ ° T(@) Pe -a) Rec>Rea>0 can be verified at once by expanding e* in powers of x. This represen- tation suggests the integration of 6,1 (2) by an integral of the form ya Sener (4 eet de, which is also suggested by Laplace’s method. Substituting this in 6,1 (2) we use the identity ae d wat pon ena~ E a) a | [ieee toe) (+ ences] aden) to show that the integral satisfies 6.1(2) provided that C does not pass through any singular point of the integrand, and that the initial and final values of e7*¢ (1+ 1)°~® are finite and equal to one another. If Re ¢> Re a> 0, one possible choice for C is the interval (0,-1), thus showing that (1) satisfies 6.1(2). If Re a> 0, another possible choice is a ray from the origin. We put 1 co jo watyont evant i -(2) ¥G, ¢; x) re de eee" (142) dt Rea>0. This equation defines a solution of 6.1(2) in the half-plane Re x > 0.256 SPECIAL FUNCTIONS 6.5 ‘The domain of definition can be extended by rotating the path of inte gration. Thus, @) VG ce [PO mttaer ae eT) “0 Rea>0, -r 0, and also the condition Re c > Re 2 > for ®, will be removed later, in 6.11(2) and 6.11 (1), when we introduce contour integrals. Another type of integral representation uses Mellin-Barnes integrals (cf. section 1.19). The formula Yy tio @ tae)° | raf Dene) fae 2m F(a) Tle +s) ~ An y>-Rea, c#0,1,2,..6, iso can be verified by evaluating the integral as the sum of residues of the integrand at the poles of [(-s). The corresponding representation of ¥ is obtained by substitmting ¥ tio Pa-c+)U4nee! > PEs) Pa-c+s+) t* ds os ym ioe 0>y>Refe —a) in (2), Tnterchanging the order of integration is permissible if y+Rea > 0, and gives P@Ta-c+0 Va, ¢52) +i “yt ive . Z dsT(-s) Pa-essen f enstyetent dy, Bai Jy~ino Evaluating the last integral, -a py tiop(. . ye Tes) Mer) Ta-ctles 4. Qmi ST ys too T@)P@-c+) or, with a slight change of notatioi, 1/7" Tass) Pe) TA~e-9) ° Wear [7 T@Ta-e+D x*ds . 3 3 ~Rea In the derivation, more stringent assumptions had to be made. ‘The exten- sion to the conditions stated in (5) follows from the theory of analytic6.6 CONFLUENT HYPERGEOMETRIC FUNCTIONS 257 continuation. The conditions imposed upon the parameters in (4) and (5) can be relaxed still further if the path of integration is suitably deformed. In fact, (4) is valid (with any y) whenever a is not zero or a negative integer, provided that the contour of integration is indented, if necessary, so as to separate the poles of '(-s) from the poles of (a +s). Similarly, (5) is valid (with any y) as long as neither @ nor a - ¢ +1 is zero ora negative integer, provided that the path of integration separates the poles of (a + s) from the poles of (~s) P(1-¢ —s). The conditions on arg x cannot be relaxed. It can be verified that (4) and (5) satisfy 6.1(2) (Whittaker and Watson, 1927, section 16,4). From (5) we have 6) Va, ¢; x) =x'~* Wa-c+1, 2-c; 2) If c is not an integer, the poles of I(-s) (1 -¢ — s) are all simple. The evaluation of (5) as the sum of residues of the integrand at these poles leads to the important relation T(-e) (2) YG, ¢3 x) eee ¢; x) Te- + T@ between the ®function and the W function. "© @(a-c+1, 2-c;x) 6.6, Elementary relations for the ¥ function The ¥ function was introduced by Tricomi (1927) who denotes it by G; it is related to F, (Meixner, 1933), to E(a, 61: x)(MacRobert 1941), and to ,F, (a, 8; x) (Erdélyi, 1939) by the relations Q) Ela, Bx) =T(@) 1 (6) x? ¥(a, a~B +154), ry Q F,(a, wane Wla,y3 x) (@) {Fy (a, Bs- Va) =x°W(a,a-B +152). (a, ¢; x) is a many-valued function of x, and we shall usually consider its principal branch in the plane cut along the negative real axis, this branch being determined by 6.5 (3) with~%r< ¢< om ‘A number of elementary relations for the ¥ function follows directly from the corresponding relations of section 64 for ®. (4) ¥@-)-Qa-c4+2)V+ala-c+ 1) ¥(a+)=0 (5) (c-a-1) ¥(c-)-(c-1L+x) +x ¥(c +) =0258 SPECIAL FUNCTIONS 6.6 (6) ¥-a¥(a+)-We-) =0 (1) (c-a@ ¥- x Wo 4) + ¥(a-) <0 (8) (a+) B+ ale—a- D ¥a4)-x¥(c +) =0 () (@-1+2) ¥-VWa-) +(a-04D) Ve) -0 (10) B' =a Vlas, c+) =¥~W(c 4) sce erteten! = Go/2) (a-0 +1) Va 4) -¥] = (1/2) [la-¢ +) ¥- Va) d ay ee Wa, cy x) =(-D" (a), W(a+n, c +05 x) a" (12) - (xe! BG, cs =D" (@-c4+D, x" Wa, c-2; x) an (13) —— [xt*""! WOe, ¢; =), (a-c +1) xo! Wla+n, o; x) dx" a a an (4) an fe7* (a, ¢; x] = G1)" e* Va, c +n; x) (15) So [em 0749-1 8a, 05 2] dx” =(-1)"e"* x68" Hla-n, c; x), 6.7, Fundamental systems of solutions of the confluent equation Four solutions, y, to y,, of the confluent equation are listed in section 6.3. From the work of section 6,5 in conjunction with section 6.3 it follows that () ¥,=¥a, ¢5 x), (Q) y,ax'6 Wa-c+1, 2~c; 2), (3) y,=e% ¥lc-a, cy ~2), (4) yy-etx" Wa, 2-5-2) are four further solutions. The relationship of thefour solutions , has been investigated in section 6.3; in particular we know that ¥ and y, ~ y, form a fundamental system of solutions of 6.1(2) provided that c is not an integer. There remains a discussion of the four solutions Y 5» ++» Yge In this section we shall assume that c is not an integer.6.7 CONFLUENT HYPERGEOMETRIC FUNCTIONS 259 The exceptional case will be dealt with in the next section. From 6.5 (6) we see that (5) ye=¥6, and y,=—e7 fT y, where € = sgn (Im x) = 1 if Im x > 0, =- lif Imx <0. This significance of « will be retained in the present section. The factor e*€7"'~*) is caused by our conventions in determining x'~*, Thus, there remain four, in general distinct, solutions, y, = ¥412=Yar ¥s=Ve andy, = ei y,, All of these are defined when c is not an integer. Their Wronskians, cf. 6,3(8), are all of the form W,, =K,, ex" with Te) ek 21 oo a. Tt) Kyneton, Kye EO gine cea fe | te 2""TG@-e+)’ In general, that is if c, a, c ~ a, are non-integer, any two of the four solutions are distinct and form a fundamental system. However, if n denotes a non-negative integer and a = ~n, W,, vanishes identically, so that y, and y, are constant multiples of each other; 6.5(7) shows that this is indeed the case. Similarly, if = 14n, y, and y, coincide; if c-as-n, y, andy, coincide; and if c-a=1+n,y, andy, are con- stant multiples of each other. If ¢ is an integer, either y, or y, fails to be defined and cannot be used. W,, has been derived, from W,, and 6.5 (7), under the assumption that c is not an integer; however, on account of continuity it holds forinteger ¢ too. Since it never vanishes identically in x, it follows that y, andy, form a fundamental system of solutions of 6,1(2) under all circumstances. The expression of W in terms of © is given by 6.5(7). The converse expression of ® in terms of ¥ is obtained by writing down an expression similar to 6,5(7) for y,, and then eliminating one of the two ®-functions, The result is Te) (ec -a) re) T(a) (1) Ba, ¢; x) = 2 £7 Wa, c; x) ns e IMO-OE @ W(e~ a,c; —2),260 SPECIAL FUNCTIONS 67 with a companion formula for y,.By means of (7) and 6,5(7), the follow- ing relations between the solutions of the confluent equation may be es- tablished: bed) TMe-D o) f@zee) (2s rae LEGA) ence 1 Te =a) : re) inae inla~e) € ey Ade ore.) (2-6) ay y,cetteramt | P@-e) | _P@~e) ee [ Td-a) 9° * Pe@-ced ry). 6.7.1, The logarithmic case The function @(a; ¢; x) is an entire fuaction of x, and also an entire function of a. Considered as a function of c, @ has poles, and fails to be defined, at c = 0,~ 1, 2, .-. - However, we have the relation ee ee eoten P(e) nal ae : showing that ®* = ©(a, c; x)/I'(c) is an entire function of both parameters as well as of the variable. In other respects too this function shows a simple behavior, for instance some of the differentiation formulas of section 6,4 become simpler when expressed in terms of ©*. As a consequence of this situation, the ® function furnishes only one solution of the confluent equation when c is an integer. fc = 1, y, and ¥, are identical. If c = 2, 3, +++, y, does not exist, and although the equation y,/T'(2 ~ c) tends to a finite limit as c approaches one of the integers > 1, (12) shows that this limit is a numerical multiple of y, and does not provide a new solution. For c = 0,- 1,-2,... , the situation is similar, except that y, and y, interchange their roles. Whenever ¢ is an integer, y, or y, provide one solution, and the second solution will con- tain logarithmic terms. This second solution can be determined by the familiar method of Frobenius. Another approach to the exceptional case uses the W function as its point of departure. The integral representation 6,5(3)defines the ¥ func- tion for all values of c and shows that this function always satisfies the confluent equation. If ¢ is not an integer, the expansion of ¥ in ascend- ing powers of x is given by 6.5(7). If ¢ is an integer it will be sufficient to assume c = L+n where n= 0, 1, 2,.... Fore =1+n, both terms on6.7.1 CONFLUENT HYPERGEOMETRIC FUNCTIONS 261 the right-hand side of 6,5(7) become infinite, and the expansion in ascending powers ofx can be obtained either by making c +n + 1 in 6.5(7), or more directly from 6.5(5). We shall briefly describe the second method. If c =n + 1, the integrand of 6,5(5) has simple poles at s =—n, —n+1,...,- land double poles at s = 0, 1, 2,..-, (ifn=0, there are only double poles). ‘The residue of (a + s) '(-s) P(-n-s) x78 at the simple pole s =r—n,r=0, 1,...,0-Lis Ot Pan tN EG alr; the residue at the double pole s =r, r=0, 1, 2, +++, is ("TP @+r) ri(n+r)! [loge + platr—yle)—yltnenl x? where o(z) is the logarithmic derivative of '(z) (cf. section 1.7). Eval- ulating the integral 6,5(5) as the sum of residues of the integrand at the poles which lie to the right of the path of integration, we obtain aye it {e1an+1:s9 tes (13) W(a, n+ 1; x) = ing tay ee) a cl r=o @+D, vy wey @ ) arnt 1 Ta 2, dene i poo The last sum is to be omitted if n = 0, The corresponding expansion of W(a, 1—n; x) can be obtained from (13) and 6.5 (6). Some formulas simplify in the logarithmic case. As an example we shall show that a a (14) fle) = (4 +2 2) G(a, c; x) da dc can be expressed in terms of confluent hypergeometric functions when c is an integer. Since Zo, =, Wa+n-vo where y¥(z) is the logarithmic derivative of the gamma function, term-by- term differentiation of 6,1(1) gives fle) 5 o Iyle+)- Ya) -2yle+m) + 2H0IZ,.262 SPECIAL FUNCTIONS 6.7.1 By comparison with (13), with n =0, (15) fC) = (2 p() = Ya) = log x] ©(a, 1; x) - (a) ¥(a, 1; x). The comesponding result for f(1 +n) with a= 0, 1, 2,..., may be written down by means of 6.4(13). 6.8. Further properties of the Y function Like the ® function, the ¥ function is a limiting case of Gauss’ hyper geometric function, since (1) lim F(a, 6; 3 1-c/x) =x* V(a,a-b +1; x) Erdélyi 19892). The proof is based on 6,5(7) and the expansion of F (1 e/) in ascending powers of The behavior of ¥ for smal] x can be investigated by means of 6,5(7), 6.713), and the corresponding formula for ¢ equal to zero or a negative integer. The results are summarized in the table W(a, c; x) for small x e ¥ | a Ee (2) |Ree>1 TP (c- Wa +R (3) |Ree<1 ell TQ-c)/Ma~c+D4R rd-c) T(e-1) =], i —— meer (4)|Ree=1, cf TaserD To" @)|e=1 =I @I loge +R See ere! ‘The order of the remainder is seen from the following table R=O(u) c x ft ©) |Rec>2 cA2 |x| Re e-2 Ty (7) |o=2 [log =| | (8) |10, and either all upper or all lower signs have to be taken. In particular, (15) A= ¥(a, ¢; ~€+ i0) - Bla, c; - €- 10) Qni oe gc al -£). Tore) Since the derivation of this formula is based on 6,.5(7), integer c are excluded in the first place. By continuity, the formula remains valid for c=1, 0,-1,-2,.... Fore =2, 3, 4, «e«, the right-hand side appears in an indeterminate form, and when it is written as Qui WE@-c+r+) T(@)P(a-c+) T(QQ-c+r) c can be made to approach 1 +n (n= 1, 2, «++ ,) with the result that Qn Tecan l+n;-é) (ar eter, r=0 (16) A=(-1)""" Pet eee We W(a, c; x) is one-valued (i) if @ = 0, —1,-2, ... , when W is a poly-264 SPECIAL FUNCTIONS 68 nomial in x and according to 6,5(7) a multiple of ®, and (ii) ife =n + 1, n= 1, 2, ..., and a is one of the integers 1, 2, ..., 7, when 6,7(12) shows that ® is a polynomial in x7'. The behavior of ¥ when x encircles the origin can be understood from the formula 17) ¥(a, 0; xe 74) e emi W(a, c; x) Ta-e) peeve ry UW) aig og; : +(-e TG aa) (a, ©; x) m0 tiptoe 6.9. Whittaker functions For some purposes it may be convenient to use Whittaker’s notation. Whittaker writes the confluent equation in the standard form 6.1(4). From 6.2(13), with a=-l, bec, a=y=0, B= we see that S(a, 03x) = e7? x8? wlk, py x) with k=-a+%c, p= %e —%. Two solutions of the Whittaker equation are the Whittaker functions (1) Mg lx) = 08 x8? Bla, 6 x= 2,, a=%-Kt+p c=2u+l. (2) Wy lz) = e787 x? Wa, 65 2) = Conversely, (3) (a, c5 2) =e x HM, (e) (4) Bla, epa)aet@ xO x) Ye-a paler tn We also have, in the notation of 6.6(3), (5) Py pede *F xX Fy (u~K +p, Wow rl /*) Further solutions of the Whittaker equation 6.1(14) can be derived from sections 6.3 and 6.7. Indicating corresponding solutions of 6.1(2) and 6,1(4) by the same subscript 6) 2H gayle Mig HA 24 ae (-x). Kyo a Mica e 2p = Mage 2 Kummer’s transformation 6,3(7) becomes (1) Mg, (2) = FON Cox)6.9.1 CONFLUENT HYPERGEOMETRIC FUNCTIONS 265 where ¢ = 1 if Im (x) > 0 and « = — 1 if Im(x) <0, and the transformation 6.5 (6) becomes 8) Wy) = Wyo. Whittaker defined M by (1), while in the definition of W he used an integral representation equivalent to 6,5 (2). Buchholz (1943 and other papers) uses the notation Kime m OX 2) = (22/a)* My yp (2) w Xz) = (2z/n)™ Wy y, (2) 6.9.1, Bessel functions If x = 0, the Whittaker equation 6,1(4) can easily be reduced to the Bessel equation 6,2(5). The result may be indicated as w(0, w=" C, Unix), In the notation of 6,1(2) this corresponds to c = 2a, The connection between Bessel functions and confluent hypergeometric functions can be summarized in the following formulas in which the standard notation is used for Bessel functions (cf. Watson, 1922). 1 (9) J (x)= Ted gx)” ee * O(% + v, 1+ 20; Qix) (+ 1 (lo) L(x) "Tor 4” e* OY 4+ v, 1+ 2v; 2x) v (LL) Mg (ix) = (2in)#** A(x) =D (ua 1) ith 224% ah J, x) (12) HO(x) = nx) ef HYT KT OP TM y Mm vs 1/(2ix)] =~ Din ete (2x) WM + v, 1+ 2r5 — Dix) =n) er ivtony | oix) (For H® , change i into ~i) (13) K, (2) = w# e7* (2x)” W(% + v, 1 + v5 2x) (14) Wo, (2) = 0% x* K 6x) (15) FG) = 274 (22)” Le OW tv, L + Qu; Bix) 4 elErMW( ay, L+2v; - ix]266 SPECIAL FUNCTIONS 6.9.1 Bessel functions occur also as limiting cases of the confluent hyper geometric functions. Performing the limiting process term by term, we find (16) lim (a, ¢; - x/a) =T(c) x4 J _, (22%), The corresponding result for the W function can be obtained from 6.5(7) for non-integral c. From Stirling’s formula, or still better, from 1.18(4), Tdee- eo aleen Gla: and this result, in conjunction with (16) and 6,5 (7), shows that (17) lim) (F(a-¢+ 1) ¥@, ¢; - x/a)] = nx¥H cosec(nc) (J,_, (22%) +e ETI, (2x¥)] sine #76 x4 He HY (2.2%) Im => 0, = ine #6 Ke HD (2504) Imx <0, Similar is the proof of (18) lim ®(a, ¢; x/a)=T(c) x4“¥e I (2x"), (19) lim [[(a-e¢ +1) W(a, 05 x/a)] = 2x4 K._, (2x), 6,9.2, Other particular confluent hypergeometric functions The relation (20) @(q, a; x) = e* is obvious, and many other special functions can be expressed in terms of the confluent hypergeometric functions. The first group of such func- tions consists of incomplete gamma functions and related functions. By 6,5 (2) and 6.5(6), and by 6,5(7) and 6,3(7), (2D Pig x)= SP ent") dese WL, 1-45 x), (22) yla, 2) =T(a)-F(, 2) For the error functions, (23) Exf(a) = fet? de = y0, 2%) = x O(1/2, 3/25-x%), x2 O(a, a4 1;-2). (24) Erfe(x)= f7e"® de = 4, x2) = e* H(%, Ms x") For the exponential integral, the logarithmic integral, integral sine and6.9.2. CONFLUENT HYPERGEOMETRIC FUNCTIONS 267 cosine and the Fresnel integrals we have (25) -EiGaje [ete dt =e* ¥(1, 1; x), pede (26) li(x) -[ ——-= Ei (log 2) = - x ¥(1, 1; — log ), ‘0 -loge (27) Si(x) = {eo sine dt =~ [07 sine de =ha~Wie* VL, L; ix) + Mie * W(1, 1; - ix), (28) Ci@e)=~ f° e7 cost de =-%e7* W(1, 1; ix) — Ke * WU, 15 a2), (29) Cw) 327% gh ts cost dt = 0% O# x [O(1/2, 3/2; ~ xe 17) + B(1/2, 3/2; - xe *)], Ha fF 1% sine de = 0% 27% 2% § [O(1/2, 8/2; — xe 17/2) - &(1/2, 8/2; - xe~ #7], (30) S@)= The parabolic cylinder functions of Chapter VIM are also confluent hypergeometric functions: (31) De) = 2% eX? WO uv, Hy Hx?) = BKPH ME WOK = iv, 9/25 Kx), Special cases of these are the Hermite polynomials: (32) H (2% x) = 2%" eX*” D (x) = 2° x WO — Mn, 8/2; Kx?) where 7 is a non-negative integer, Again ¥(, 0;*) = x W(a + 1, 2; x) is connected with Batemans function k, (4x) with v = 4 - 4a. Originally Bateman put (38) k(x) = 207" f°" cos (& tan 0-6) 0 for real values of x and v. Then (34) T(v + D hy, (2) = FW vy 05 2x) for x>0, and it is useful to consider this as the definition of the k-function in the cut plane. With Bateman’s original definition, &,(-x) = k_,(x). This formula no longer holds when the definition (34) is adopted. Instead we have (38) ky (-E +0) =k_y(€)— et e% Ov, G - 26) é>0.268 SPECIAL FUNCTIONS 6.9.2 If a is zero or a negative integer, ® and Ware polynomials in x, related to the generalized Laguerre polynomials of Chapter X, (a+ -)" (36) Ee), tial gn, a+); py sae Ben, a+; 2) - 7 see extaha! 0s ee: The so-called Laguerre functions, for unrestricted values of y, are an altemative notation for confluent hypergeometric functions (Pinney 1946) (37) L@ (= _o¢ 1x) x) = Hy athe z Pee aa If ¢ ~a~1 is a non-negative integer, x°~' W(a, c; x) is a polynomial and (38) P(a, a+n 41; x)=27°" Ban, i-a-n; x) = (a), 27°" O(-n, 1- an; x) = (a), x PLE (x) 2=0,1, 2,555 It can be proved (Erdélyi 1937f) that the confluent equation has a solution which is a finite combination of elementary functions if and only if either @ or c —a is an integer. Of recent importance are the Poisson-Charlier polynomials (Szegé 1939) arising in the calculus of probability. They can be expressed by means of the confluent hypergeometric function as (39) p(x) =a7* a) (w-2 4D, On, x~n+ ha) The so-called Toronto functions (Heatley 1943) are defined by 2 [Gm+h) T(1+n) The following table gives information about some of the more important particular and limiting cases of confluent hypergeometric functions. (40) T(m, n, x)= x27B th em# b(ymt+ n+l; x). SPECIAL CASES OF THE CONFLUENT HYPERGEOMETRIC FUNCTION Parameters Functions G21, 2.665 Incomplete garama functions and their particular cases a=0,~ , Laguerre polynomials are Bessel functions o=0, cH? Bateman’s k-function c=, c= 3/2 Parabolic cylinder functions cna Elementary functions, incomplete gamma functions6.10 CONFLUENT HYPERGEOMETRIC FUNCTIONS 269 SPECIAL CASES OF THE CONFLUENT HYPERGEOMETRIC FUNCTION (Continued) Parameters Functions @ Laguerre polynomials, e-a=1,2, 3,005 W Incomplete gamma functions c=2a Bessel functions exa=1 Exponential integral and related functions a=1/2, ¢=3/2 Etzor functions and related functions 6.10. Laplace transforms and confluent hypergeometric functions We shall use the notation (1) LIF(t) s}= ee F(t) dt = f(s) for the Laplace transform of F(t), and shall occasionally write more briefly L1F}. From the theory of the Laplace transformation we quote the product theorem, 2) LAS! Fw) F(¢=) dul=LU FO} LF, OS, valid, for instance, if L1F,} and L1F,} converge absolutely, and the complex inversion formula 1 opr tie (3) FO -af f(s) ettds 2m J, — ino valid, e.g., if L{F § is absolutely convergent for Re s = y, and F(x) is of bounded variation and continuous in some neighborhood of ¢. The, infinite integral is in general a Cauchy Principal Value, that is limi oe us ice. Hogue, if Ec) is absolutely integrable on Re s = y, we may write simply cee ‘There are many pairs of Laplace transforms in which confluent hyper- geometric functions occur. We may rewrite 6,5(2) as (4) Lae" 4} =P @) Bla, ¢; x) Rea>0, Res>0, We also have, for Re 6 > 0, and Re s > Max(0, Re &), (5) Lit?" B(a, c; ke)} = 16) 5° F (a, b; 5 ks™") Is] > [A], =T(6) (s -2)7* Fle ~a, bs cs kk ~ 5)] Is-2] > fal. The first form can be obtained either by term by term integration of the270 SPECIAL FUNCTIONS 6.10 power series for ©, or from 6,5(1) and Euler’s integral for Gauss’ series; the second form follows by Euler’s transformation of the first form. If k is negative real, we may make s + 0 in (5) provided that Re a> Re 6. With b=c, (5) assumes the simpler form 6) Lie*™" ®(a, ¢; )}=T(c) s"¢(1-s™")-8 Rec>0, Res>L From (5), 6.5(7), and 2,9(33) () Lie" W(a, 5 eh P@)T(b-c+) — ~c+]; -c+l1-s Teka F(b, b-c+la+b-c+]1-s") Reb>0, Rec 0. If Re a> Re 6, we may make s » +0 in (7). Another Laplace transform pair is given by Weber’s first exponential integral ~' F(a, bja+6-c+l;1-s") Res>h T(a) T+ (8) Lies" (20%) b= “* O(a, v41;-s7') Rea>0, Res>0. Similarly * (9) Lies" K (22) =% Pa) Pav) 8? (a, v4.45 574) Rea>0, Rea>Rev, Res>0, The relation (10) Lte®" Ffa, 6; es -O)}=T(c) 54° * ¥(a, a— 5 +1; s) Rec>0, Res>0O is equivalent to an integral representation of W,., given by C. S. Meijer. Another transform pair is (11) Lte7# £7? O(a, 3 #2) = 2'7 F(2e- Y) We -¥, a + Hs Ks*) Rec>%, Res>0, In Whittaker’s notation, the principal formulas read: (1D LheMe2M, O}= Tas p+ 9/2 se? x Flat pt 3/2, p—x + 1/2; 2p +1; 87") Re(a+y+3/2)>0, Res>Q6.AL1 CONFLUENT HYPERGEOMETRIC FUNCTIONS 271 (13) LteAe#"# a (b= Tp 1) (8 = A= WYP (sd OHH Rey>-%, Res>RerA-%, (14) Liew, (0) Pat pt 3/2) Ma~p+3/2) Tta=«+2) x Flat+p+ 3/2, p- wx 41/23 a-K +2; 1-58) Re(atp+3/2)>0, Res>0. From the product theorem, (2), and (6) we have the integral addition theorem - _ elt 0 fF 5 Bo) oe TOM Te’) pete“ “Tleted The particular case a’ =0, of (16) fu = WI OG, ys u)du spe eee! Masa’, ct+e';2) Rec>0, Rec’>0, LO Me=W) 564 oa) Rec> Re y>0 T(c) may be mentioned separately. 6.11. Integral representations In this section the basic integral representations will be generalized by the introduction of contour integrals, and further integral representa tions will be listed. 6.11.1. The © function ‘The integral reptesentation 6,5(1) is based on Euler’s integral of the first kind, 1,5(1), for the beta function, The restrictions imposed upon the parameters can be removed, partly or entirely, by the introduction of complex integrals, as in section 1.6, Using 1.6(6), 1.6(7), ot 1.6(8), instead of 1,5(1), we arrive at the integral representations (1) O(a ¢; x) =m) ee P(1- a) Mo) P+) xf MPP IED pat part Cy pervert gy272 SPECIAL FUNCTIONS 6.111 (2) (a, ¢; x) = (Qn) P(e) Fla-~c+ D/T(@ xSP emt eG ee ae Re a>0, and (3) @(a, 65 2) == Qn)! P(e) FA -a)/F(e - a) xf onda yerant ae Re(c =a) >0. In (1), the contour of integration is a double loop starting at a point A between 0 and | on the real ¢ axis, with arg ¢ = arg(1 - 2) = 6 at A, en- circling first = 1 in the positive sense, then t = 0 in the positive sense, then t = 1 in the negative sense, and finally ¢ = 0 in the negative sense, returning to A. In (2), the contour is a loop starting (and ending) at ¢ = 0 and encir cling 1 once in the positive sense. Similarly in (3), with 0 and 1 inter changing roles. All powers have their principal values in (2) and (3). SDP (1+ 0+ 1-, 0-) The function can also be represented in terms of Bessel functions. From 6,10(16) with y = 2a, and 6,9(10), (4) P(a) P(e — 2a) ®(a, 5 x) = 7% P(e) x¥* «SL e Kxt p0™K (1) _ p)enmzo-t Ty, (at) dt Rec>2, Rea>0, and from 6,10(8), with a slight change in notation, (8) T(c~a) ®(a, ¢; x) =T(c) e* x ¥ He x J ev tetere hk J, (2Caa)¥] dt Rec>Rea>0, Rex>@ Further integral representations may be obtained by using the complex inversion formula, 6.10(3), in connection with the Laplace transforms 6.10(5), with k = 1, and 6,10(6). (©) (a, ¢5 2) =(2ni)" PB) e8-> hes efts"8F (a, b; 05 s-") ds Reb>0, y>1,6.11.2 CONFLUENT HYPERGEOMETRIC FUNCTIONS 273 (1) BOG, 65 ) = (2m P(e) et 8 [74 ettsme (1-8) ds yo ie Rec>0, y>ol Ifb=n+1,2=0, 1, se, in (6), the integrand is a one-valued function of s, and the path of integration may be replaced by a closed contour, for instance, by a circle |s| = p> 1 (8) (a, c; 2) =(2mi)"! nites ets"! F(a, n+1;¢;s~') ds 2=0, 1,2, e008 6.11.2. The ¥ function A similar discussion of the W function leads to the following integral representations: to#) (9) W(a, ¢5 deen emra-a f o ee" (La e)ee de oe -Mn0, Re(a-c)>~1, (11) TP) ¥le, e5x) =x) [Mento F(a, a—c +1; bs—t) dt ° Reb>0, Rex>O, (12) Tla+b-c4 1) Wa, ¢; ) = (2m) "TD Pe-c+ etme x $77 etts-0 F(a, bsatb—o+]; l-s")ds y= 0 Re b>0, Re(b-c)>-1, y>% For x > 0, the following integral representation can be derived from 6.5(2), We assume Re a > 0, Re c < 1, x > 0, Then it is permissible to deform the path of integration into the segment of the rea] axis from t = 0 to t= — % and the ray fromt = — % to t=—%+ ies, Along (0,-4)we putt = ue *”; along (-4, ~ % +i) we put tee sec8, l+t=he sec O(0< OS %nh Thus we find 27 P(a) (a, 03 2) = fe we (L~ ue“ du % ~i2'-* f°" (cos 0)" exp [4x - Mix tan 0+ (c - 2a) 161 d6.274 SPECIAL FUNCTIONS 6.11.2 A corresponding formula can be obtained when the path of integration is deformed into the segment (0, -%), along which now t = ue™*”, and the ray ¢}, -% — ix), along which ¢ = %e 9-7) sec 9, Subtracting the two formulas, (13) 7 Bla, ¢; x) = 2! (1- a) e%* x ce (cos 6)* cos(%x tan 0+ (2a—c) 0] d0 a Os Rec 1, ami occa The condition Re a > 0, used in the derivation, can be discarded by analytic continuation. However, a new restriction on @ must be introduced in order to avoid the poles of the gamma function. Formula (13) corre sponds to the integral representation 6,9(29) of Bateman’s & function. Two further integral representations, due to C. S. Meijer (1938, 1941), are (14) W(a, ©; x) = x97# f° “Ke cosh 2¢ Pu (cosh ¢) (2 sinh t)'# de p=c-2a+%, v=c-3/2, Rex>0, Red<]l, and (15) Wa, ¢5 x) = 778 2e-Hett oe gh Ke xf eH ih Fp .. [(2x)* cosh t] cosh [e — 1) #] dt Re x>0. 6.11.3, Whittaker functions The basic integral representations for Whittaker’s confluent hypergeo- metric functions are: (16) POS + WTO +R +H) My) = 2° D(2y 4 1) 2th i MAE] APH CL 4 WAR tH dy Re(u tx) >-%, 17) TH+ K+ My, = T(Qpt DeMx¥f en te# J, [2(tx) 4] de Re(p+x)>- (18) PO - K+ WP, mene hth font KKH] 4 AH HK de ° Re(u-x)>-, Rex>0,6.12 CONFLUENT HYPERGEOMETRIC FUNCTIONS 275 and 19) Wg (2) = (2nd)! eH a ef EO Rinne Tatum) ASS Ae EG ete) Th-«-pTU-K+p * : "yrs mex< oa, where neither 4+ « +p nor 4+ — pis a positive integer, and the path of integration separates the poles of IT (s) from the poles of TQg-«-p-s)TK-K+yp-s). 6.12. Expansions in terms of Laguerre polynomials and Bessel functions Beside the power series expansions, there are other useful expansions of confluent hypergeometric functions. We put =9, c= a+1, ¢=u/(1-x), in 6.11(9) and obtain Q) Wa a+]; x) = (ni) “he? TA -a) x fen ermatma ue" (1 u)72" du. Now, 2) (Le a)tet emai Bp) yn {ul <1 ° is the well-kaown generating function of generalized Laguerre polynomials [Szegs 1939(5.1.9)]. Since this series converges only inside the unit circle |u| <1, we write f°" = lim f°", v + 1-, and substitute (2). Since « ' Se Putte! du = Qie * sin (an) vn +), we thus obtain TC) Wa, a4 Ls 2)= Jim ot Enea tor LOG) pet~ nso and by Abel’s theorem on the continuity of power series, (3) PCa) (a, a+ 1; 9) E (sat (x ° whenever the infiaite series converges. From the known estimate of generalized Laguerre polynomials (Szegé 1939 Theorem 7,6.4) it is seen that the series is convergent in any finite positive interval of x if a < 4. Cn the other hand, using the transformation 6.5 (6),276 SPECIAL FUNCTIONS 6.12 (4) Ta-a) Va attyxan% 2 (nea-a) LO") nee where the series converges (for positive x) if a> — % Outside the pos- itive real axis both series are divergent. The expansions (2) and (4) are due to Tricomi. Another expansion in terms of Laguerre polynomials (8) 6, es ay(z-)I=-2* S {, nao (), LEG) 9 le]<1, y>9 is a generalization of (2) and, with x = }4, it provides an expansion of © on the negative real axis in terms of Laguerre polynomials. It is due to A. Erdélyi [1937 a, equation (5, 7)]. Tricomi (1947, 1949) has given two expansions of © in terms of Bessel functions. These expansions are useful for the investigation of the be- havior of @ when the parameters are large. The first expansion is (6) Oa, a+1; x) =T(a+t I) (ax) *%e™ z. A, i) @/a)** J, [2(ax)*] where @ is real, h > 0, and the coefficients A, are determined by the generating function @[L+(h—-Dz]*(+hzy ret. Gy) re 2 =o From the asymptotic representation of Bessel functions for large order it is seen that the infinite series in (6) converges as i A,(h) x"/T(atm+1) and hence is absolutely and uniformly convergent in every bounded region of the complex x plane, From (7) we have (8) Aj=1, Aye-le+ Dh, Apa h—-W at¥ (atl) (a+ 2) A? and the recurrence relation (9) (m+ A, =[(1~2h) m-Alat DIA, -[(1- 24) +hQh~DYlat mld, thh-Yad,, 2 =2,3, 4,000, for the computation of the later A. Also, (10) 4, @)=C)* Le" a) A, (1) lee tot), and ifa=n=0, 1, 2,..., also6.13 CONFLUENT HYPERGEOMETRIC FUNCTIONS 277 4, (i) =ae a = Pk LTO FOE Lah), ao The A are polynomials in a, a, h, and the degree of A, in a is [4m] if h#£%, and [m/3]ifh = ¥ ((B} is the largest integer <4)- The most suit- able choice for h is 4. Tricomi’s second expansion is (11) e™% @(a, a+ 1; 2) = Plat LD) (xx) x EA gles B+ Yad hu ay" I. (On% 2X) where x =} + %a— a is Whittaker’s parameter, and the A. (x, A) are coefficients in the expansion (12) eX (1= 2A (14 JHE x, do lel « can be investigated by means of the basic integral representations. If in 6.5(3) we put GQsnret= F Cy @-c+D, 2) and estimate the integral involving Ry , it can be shown that @) Wa cade ¥ Ce Malem et De pooen giie|e¥nt no nt 3 3 N=0,1, 2,006, *%20, -=a< 0. The corresponding result for ¢ + «© when ¢ - a and x remain bounded follows from 6.5 (6). The case a + « is more difficult. Perron (1921) based his investiga- tions of this case on integral representations of the Laplace type and used the method of steepest descent. Among later investigators we men- tion F, Tricomi (1947), who showed that under certain restrictions his expansion 6,12(11) is an asymptotic expansion, and W. C, Taylor (1939) who used E, R. Langer’s method of asymptotic representation of solu- tions of differential equations. Taylor’sresults are described in 6.13.3, If |x| is bounded, and bounded away from zero, they simplify consider- ably. In the formulas below, c is bounded, x is bounded and also bounded away from zero, @ + =, It will be more convenient to express the formulas in terms of () Kake-a It is assumed that larg x - arg x/ <7. We then have the following table in which ¢ is any positive number: KA mK gh tHe eM Wa, os x) x Scratel arg x between 2% cos{am — 2(xx)% — Ya} - (1+ 0([x|-)] te, mHe€ U(14i)exp(-ixm+2i(xx)*1+ (140 (l«|-9)) 42% cos[xn—2 (kx) 4+ a) [140 ({x{ 7401 = (=i) exp [icm—2i (xx) *] «(140 ([K|74)] 3a-e€ (9) (10)] 27 +e, 6 4n-e an] 3e+ 6 One The asymptotic form of ®, under the same circumstances is obtained from 6.7 (7): a (12) @ (a, 5 x) = 24 P(e) oF MME ME Le, RHE 4g eWBLiK a! + (cx) 0 [exp [Im (2«% «|B280 SPECIAL FUNCTIONS 6.18.2 and with s an integer we have (13) ¢, = (2n)7% ¢ ime) e=1) 1 Qs-Datesarglex)* 0, The case where a, c, and c — a > simultaneously, has not been investigated fully, It is, however, known that if (16) a=vA+a, c-a=vB+B where a, , are fixed, possibly complex, numbers, A, B, are fixed positive numbers, and v > o through positive values, and if the abbre- viations t = - A/(A +B), u= A(1 +t) = AB/(A + B) are used, then (2n)* P(e) @y* T@P(c-a) See also section 6.13.3. (17) (a, c;x) = e7(-2)*(1+2)°*[1 +0 (YY 6.13.3. Variable and parameters large Hf @ is bounded and c and x » « in such a manner that (x| <|cl, the behavior of ® may be investigated by means of 6.1(1) We put x=c€,0<|€| <1-e(e>0), and use 1,18(4). Then we have as c > 0 1 Ft) ©, Tern =e™™{1— Yn(n- 1c“! + O(|e|~)} and : (18) (a, evef-0-0~[1 022 (2) ote abounded, |é|< l-¢, ©>0. The corresponding formula for bounded c~a, andc, x + « may be obtained by means of Kummer’s transformation, the formula for ¥ by means of 6,5(7). When a and x increase indefinitely, the behavior of confluent hyper- geometric functions is more complicated. W. C. Taylor (1939) has applied E. R, Langer’s method to derive the asymptotic forms of con- fluent hypergeometric functions from the differential equation. We shall use again, (19) k=%e-a.6.13.3 CONFLUENT HYPERGEOMETRIC FUNCTIONS 281 ‘Taylor introduces the auxiliary variable (20) E= ita lx — 4) —« log [Ox + (x - 4x)*)? 40), where the arguments of x, x, and x — 4x are all zero when these quantities are positive, and for other values the arguments are obtained by analytic continuation in such a manner that |arg x ~ arg «| < 7 throughout the process, First Taylor investigates the asymptotic behavior of W under the assumption that £- co and that there are constants r and N such that O00, and |x| > N|x|~'*" as x > 0, His results for this case are: arg € between OK OK (Ag) % eH Wha, 03 x) = Qb|-27+64 —e (eo ~ ie7 FF) (140 ([a| “4 0((E I (22)|-7+e, 2n-e€ e [14 Ox] + OE] (23))r+e, 30-€ (e* 4 ce~ YT + O(\x|~) + OCE!"DI (2M) 2r+e Sa-e ie 11 + Ola) + OE | I For bounded &, or x ~ 4 =O (|x|'), Taylor has the asymptotic form (25) 7K 4“ (xe — Ax) eX Wa, 05 x) =[2 7 E/3A)I [e278 I, pg (E) ~ 7 #77 Jay (2) + O(n") 2-4 =O(|K\"%), Kr 0% For ®, Taylor has (26) (xx)#e~* e“#* Bla, ¢; x) =D Ce) F,_, [2¢ex) 4] + «7% 29/4 Ofexp |lim(2x% x*)|] cy argx, arg x bounded, x=O(\x|'279), xo 00, If xx is large, the Bessel function can be expressed in terms of elemen- tary functions Tn x) = 2% HK a Le, erin +e, evzitkalt + (xx)7* Ofexp [Im (2x x’) where ¢, andc, are as in (13) and (14). Another result for large @ and large x is that Tricomi’s expansion 6.12(10) provides an asymptotic representation if x > « and x = O(|x|*) with some p < 1/3 (Tricomi 1949). The case when both parameters and the variable are large has not been investigated systematically, Erdélyi (1938d) applies the method of steepest descents to the integral representations of the Laplace type to discuss the behavior of confluent hypergeometric functions when282 SPECIAL FUNCTIONS 6.13.3 (27) x=vX+8 anvAta c-a=wB+B 4, 8 X real, and either (28) 4>0, X>0 or (29) A>0, B>0. The real numbers 4, B, X, and the possibly complex numbers a, 8, & are fixed while v +o through positive real values. The quadratic equation int, (30) Xe(¢+ 1) -A(e+ D-Be =0 has two distinct real roots in both cases considered. In the case (28), Jet 2, be the (only) positive root; in the case (29) let ¢, be the (only) root between ~ 1 and 0 of (30). Also put (31) u,= ACL +¢,)? + Bey = (L+2,) (A + Xe}) hai,2 which is positive in both cases, Then we have (32) D(a) Bla, 63 x) ~(2n/u, e181 40,881 4 O07 NI A>0, X>0, vse and (33) T(@) P(e - a) ®(a, ¢; x) =D (c) (2a/u,)* e 2 (He.)2 (140,411 4 007) A>0, B>0, vox which can be proved from 6.5 (2) and 6.5(1), respectively. 6.14, Multiplication theorems a Aa 1h x” a" ia) 2aaee fe ° nt dx" on the one hand, and the expansion AfQx) = ae ,, — 1a = feo, which is a particular case of Lagrange’s expansion (E, T. Whittaker and G. N. Watson 1927, section 7.32), on the other hand are sources of the “multiplication theorems” for functions for which either all d” f (x)/dx" or all d* [x f(x)I/dx" are known. Using the formulas of section 6.4 and section 6,6, we thus arrive at the following multiplication formulas:6.15.1 CONFLUENT HYPERGEOMETRIC FUNCTIONS 283 () @(c;ax)= ) Oe QQ=- D2" Blasn, c+n; x), acto WI (ec), (2) @(a, c; Az) = ate 3, ore. a. 2)" @(a, ¢~n; x), - (3) ®la, cs Ad =27* On (1-A7)" O(a42, ¢; =) Rer>%, fon! ) WG ers= On (yeyre" Worn cen) lA-1<1, nt Ko (5) Wa csasy=atre Y eres Pa ye a)" Wa, © =n x) 2a ee (6) Wa, ¢; Ax) = a7 2 @a =e Dn yy wie sn, c2) nl |A-I <2, Ready All these formulas can he re-written as addition formulas by putting As L+y/s, Ax= ay. A further multiplication formula, (2) ®G, ¢ aw J Tahar COT Fn ot ») Ho x @(a+n, g+2n41; x), was given by Erdélyi (1936c). Here g is anarbitrary parameter except that it must not be a negative odd integer. The Gauss series F appearing in (7) is a Jacobi polynomial, and becomes an ultraspherical polynomial when g = 2c — 2, and a Legendre polynomial when g = c = 1. 6.15. Series and integral formulas A vast number of relationships involving infinite series or integrals of confluent hypergeometric functions is to be found in papers of the iast 20 years, No unified theory exists, and a full presentation of the results is impracticable, Only samples of the more interesting results and some references to the literature will be given. The references are far from complete, and further papers will be found, in particular, in Eng- lish and Indian periodicals. 6.15.1. Series Many of the series of confluent hypergeometric functions which were investigated have one of the following three forms:284 SPECIAL FUNCTIONS 6.15.1 = a, (an, ox) a=0 QQ) EF Bix" @(atn,c+2n5x) mo (3) © yx" @(ac+n; x) neo A few of the results are showa in the table below | Coefficients Sum =(c-c'),/n! r fo | er AO 6e 6 94, 2) Re@e'- 0) >% re) a,2tene'),/nt | Pee IP (e~e N= ee xt (5) | Rec>Ree!>0, x fytut “ e-uyee™ @(a, 05 uw) \t}<2, largx|Re v>0. x O(a, cv; x -u) du For these and other, related, results see Erdélyi (1936b, c, 1937a, c). In 6); (9) 4,02 2 40" Fem, a~c-n+1;2-c)/m}. sto For other series see section 6.15.3. 6.15.2 Integrals Indefinite integrals with confluent hypergeometric functions follow from the differentiation formulas of sections 6.4 and 6.6. Many definite integrals can be derived from the formulas of section 6,10. If k =-1 and Re a > Re 6 > 0 in 6,10(5) or if Rea > Re 6 > 0 and Re c > Re b + 1 in 6,10(7), we may make s + 0. Thus6.15.2 CONFLUENT HYPERGEOMETRIC FUNCTIONS 285 1) Me) (a= d) 20 bt har (10) _ 28" @(a, c3 = 2) dt TiaTFe-b 02 Re(e-2a)<% (which is the Hankel inversion of 6,15(19)], the reciprocity formula (14) Pea) SP 08 14 ne" WOa, 05 tx) dt =M(a) f° ert (14 ee" Was os tx) de ° Rea>0, Rea>Rec'-1, Rea’>0, Rea’ >Rec-1, Magnus’s addition theorem (1946) (15) (ni)! f ** Pea) P(e - 2) W(a, ¢; x) We - a, ¢; y) da = =T(c) ¥(c, 2c; x +y), and the formula (16) fo "#28" (x 4 97? O(a, 0; x) dx =(-D" P(e) F(a) y**""! We - a, 5 y) -Rec0,286 SPECIAL FUNCTIONS 6.15.2 =(a/€) e~£ [@(a +1, ¢; €)} é=7, Rec>0, and for any two zeros é, 7, of ¥ SP" Dale — K(E+ eK E40" ce Wa, c5 Ex) Wa, esx) dx = 0 é#n, ao ef Wa, OP | We also mention here an inversion formula with the kernel N(k, x) =e" *? Ole + ik, 5 ix) e>0 From f(z) = S" NG, 2) g(h) dk if follows, under certain assumptions that The +i) Phe th) gy pe ony oa IF@l eft f yet NUks ~) £0) aye 6.15.3. Products of confluent hypergeometric functions glk)= The investigation of products of confluent hypergeometric functions often involves generalizations of the hypergeometric series (cf. Section 5). In this section we note some of the cases in which such general- izations do not occur. Some of the more important integral representations are: (17) e7#7¥7 @(a, 65 x) Blas 65 x) = (ni) De) ff et(s Mas By) st hx—hy)@ x(s+hx 4 Kyte’ Fla, a'sc; 4zy[4s2=(x-y)2I"} ds where L is a loop starting and ending at — , and encircling all singu- larities of the integrand, i.e., the four points s = + 4x +%y, in the pos- itive direction, (18) T(@) P(e - a) ®(a, ¢; x) ®(a, ©; =z) = (P(e)? x1 fT, @ sech 2) e'%* sech ¢ de Rea>0, Re(c-a)>0, (19) F(a) ®(a, 65 -x) W(a, ¢; x) = P(e) xt f” J_, ( sinh ¢) (tanh Ye)" de Rea>0, Rex>0, (20) 7 Wa, ¢; x) W(e~a, 3 x) = 2x * et [47 K,_, (& sect) cos[(e - 2a) t] sect de Rex>0,6.15.3 CONFLUENT HYPERGEOMETRIC FUNCTIONS 287 (21) FQ) Va, 5 2) Bas, es y) =f ete (na rey sgn ° xFla aby t(ety td (e+e Geel de y=ata'=c+1, Rey>0, x0 We may add the following integral formulas: (22) Sen#1e"! @(a, 0; 2) @ (aso; At) dt =T(c)(s-D7*(s-N-*" sete Fla, a0; Ms-1)7(s-A)~*] Rec>0, Res>Redr+l, (28) Dasa’) JP y¥ee Ts (2G) *] Wa, 05 9) @ (a! 245 ~y) dy =D (ce!) 2¥etke'"1 We! a! c+ 0! ana’; x)O(a',a 405-2) Rec’>0, 1 (lc), at = (a), (a’) sey eT Metm etm en 2) Meline’4my—2), ako288 SPECIAL FUNCTIONS 6.15.3 2) Y POW e gasnc tmx 2) O(a c/n: ye" sO, © Vqnt Sy: . > ee ee (0), €),a! ae YO en ¥ ey, POLE N= n= 0 & & =a-3%) aca Lh 4n, © +15 x2(2 - D1) x O@[h +n, o! + m5 yz(z — 7} [xyz (1 - 2)77)" le 0 or a—c+1>0. If -n 0 are govemed by 6.8(2) to 6.8(5). The information derived therefrom is represented in the diagram below. The results coincide with those by A, Milne (1915) and G. E. Tsvetkoff (1941a). Positive zeros of W(a, ¢; x) Approximate expressions for the zeros have been given by Tricomi (1947). From 6,12(11), it can be shown that if €, is the r-th positive zero of O(a, c; x), and j,_,, the rth positive zero of J,_, (x), then for large x, ] 4x73 40K). 17 QM) €=/2,, G0" 11+ 3c le - 2) + jf H. Schmidt (1938) found a similar result and showed that the series of which (1) gives the first two terms is convergent for sufficiently large a|. The r-th positive zero can be approximated by (2) (r+ %e-¥)?/(2e - 4a), Further details about the zeros are contained in the papers by Tricomi and Tsvetkoff quoted above. 2 emu6.17 CONFLUENT HYPERGEOMETRIC FUNCTIONS 291 The complex zeros (for real a and c) have been investigated by Tsvetkoff (1941 b) and by Tricomi (1950a). 6.17. Descriptive properties for real a,c, x ‘The results of section 6.16 together with differentiation formulas such as 6.4(10) and 6,6(11) give information about the number and approximate position of the zeros, turning points, and points of inflexion of the con- fluent hypergeometric functions when a, c, x, and hence also © and ¥ are real, Moreover, the Sonine-Pélya theorem (Szegd 1939) gives results on the magnitude of successive maxima and minima, Writing 6,1(2) in the self-adjoint form ay fares) ae tenty a0, de dz 278 an application of that theorem shows that the tuming values, or rather their absolute values, form an increasing or decreasing sequence accord- ing as (2) axe"! ex 8 et = an 2en! en is a decreasing or increasing function of x, Hence the successivemaxima of |y| are increasing if (3) a>0, xc-%, and are decreasing if (4) @>0, x>¢e-% or a<0, x 0 and x is outside of 5 (5)4 or if x< 0 and x is inside 5, and are decreasing if «> 0 and x is inside 5 (6)< or if «<0 and x is outside 5, The sizes of the later tuming values can be approximated by means of the asymptotic representations developed in section 6.13. As an example, we shall investigate () y= O(- 45, 1; x)292 SPECIAL FUNCTIONS 6.17 for real x. In Whittaker’s notation, x = 5, 1 = 0. By 6,4(10), (8) y'=-4+5 &(-3-5,252). Clearly y (0) = 1, y‘(0) = ~ 4:5, and since P' (- 4+5) < 0,we have from 6413(3) that y(~%) = 0, y'(- =) = 0, y(~o) == 0, y! (x) =— oo, From the diagrams of section 6,16, y has five positive, and no negative zeros. From 6,16 (2) the zeros of y can be approximated by 0.3, 1.5, 3.7, 6.9, 10.6, from 6,16 (2) and (8) the turning pointsby x = 0.9, 2.8, 5.8, 9.9, Moreover, at all these points (2) is satisfied and so the maxima of |y| form an in- creasing sequence. A rough graph of y based on this information is given below: OC 48, 15x)CONFLUENT HYPERGEOMETRIC FUNCTIONS 293 REFERENCES Appell, Paul and J. Kampé de Fériet, 1926: Fonctions hypergedmetriques et hyperspheriques. Polynomes d’ Hermite, Gauthier-Villars. Airey, J. R. and H. A. Webb, 1918: Philos. Mag. (6) 36, 129-141. Archibald, W. J., 1938: Philos. Mag. (7) 26, 415-419. Bailey, W. N., 1987: Quarts J. Math., Oxford Ser. 8 51-53. Bateman, Harry, 1931: Trans. Amer. Math, Soc. 33, 817-831. Bock, P. L., 1939: Compositio Math. 7, 123-134. Buchholz, Herbert, 1943: Z, Angeus Math. Mech, 23, 47-58 and 101-118. Buchholz, Herbert, 1947: Z. Physik, 124, 196-218, Buchholz, Herbert, 1948: Ann. Physik (6) 2, 185-210. Buchholz, Herbert, 1949: Math, Z, 52, 355-383, Burchnall, J. L. and T. W. Chaundy, 1940: Quart, J. of Math, 11, 249-270. Burchnall, J. L. and T, W. Chaundy, 1941: Quart, J, of Math, 12, 112-128, Cailler, Charles, 1920: Enseignement Math. 21, 224295 and 255-259. Chappell, G. E., 1924/25: Proc. Edinburgh Math. Soc. 43, 117-130. Dhar, S. C., 1933: Bull, Calcutta Math, Soc. 26, 57-64. Erdélyi, Arthur,1936a: Math Ann, 118, 357-362. Erdélyi, Arthur, 1936b: Nederl. Akad. Wetensch., Proc. 39, 1092-1098. Erdélyi, Arthur, 1936: Monatsh, Math, Phys. 45, 31-52. Erdélyi, Arthur, 1936 d: Math. Z. 42, 125-143. Erddlyi, Arthur, 1937 a: Math, Z, 42, 125-143 and 641-670, Erdélyi, Arthur, 1937b; Akad. Wies, Wien. S-B Ila 146, 431-467. Erdélyi, Arthur, 1937¢: Monatsh. Math, Phys. 45, 251-279, Erdélyi, Arthur, 1937d: Quart. J. Math Oxford Ser. 8, 200-213. Erdélyi, Arthur, 1937e: Monatsh, Math, Phys. 46, 132-156. Erdélyi, Arthur, 1937{: Monatsh, Math. Phys. 46, 1-9. Erdélyi, Arthur, 1938a: J. London Math, Soc. 13, 146-154, Erdélyi, Arthur, 1938b: Philos. Mag. (7) 26, 871-877. Erdélyi, Arthur, 19384: Casopis Pest. Mat. Fys. 67, 240-248. Erdélyi, Arthur, 1938c: Nederl, Akad. Wetensch,,Proc. 41, 481-486. Erdélyi, Arthur, 1939.a: Compositio Math, 7, 340-352, and 6, 336-347. Erdélyi, Arthur, 1939b: Quart, J. Math, Oxford Ser. 10, 176-189, Erdélyi, Arthur, 1939c: Proc. Benares Math, Soc, 1, 39-53.294 SPECIAL FUNCTIONS REFERENCES Erdélyi, Arthur, 19394: J, Indian Math, Soc. 3, 169-161, Erdélyi, Arthur, 1940: Nieuw Arch, Wiskunde 20, 1-38. Erdélyi, Arthur, 1941: Proc. Roy. Soc. Edinburgh 61, 61-70. Fisher, E., 1935: Proc. Nat. Acad, Sei. U.S.A, 21, 529533. Gheorghiu, G, T., 1938: Bull Sei. Ecole Polytech. Timisoara 8, 17-27. Giraud, Georges, 1942: C. Re Acad, Sci, Paris, 214, 649-651. Gordon, W., 1928 Ann. Physik, 48,180. Gran Olsson, R., 1937: Ing. Arch. 8, 99-103. Gupta, H. C., 1943: Proc. Nat. Acad. Sei. India Sect, As 13, 225-231, Heatley, A. H., 1943: Trans. Roy. Soc. Canada (3), 37, 13-29. Howell, W, Js, 1939: Philos. Mag. (7) 28, 493-495. Humbert, Pierre, 1921: C, R, Acad Sci. Paris, 173, 217-219. Kienast, Alfred, 1921: Denkschr. Schus Naturf. Gesell. 57, 247-325. Kummer, E. E., 1837: J. Reine Angew. Math. 17, 228-242, . Math, Phys. Moss, inst Tech. 21, 264 Lowan, Amold and W. Horenstein, 1942: cee exreku MacRobert, JX M., 1919/20: Proc. Roy. Soc. Edinburgh 42, 89-96. MacRobert, K. M., 1938: Philos. Mag. (7) 26, 82-93. MacRobert, R. M., 1941: Philos. Mag, (7) 26, 82-93. Magnus, Wilhelm, 1941: Z. Physik 48, 343-356, Magnus, Wilhelm, 1946: Nachr. Ges. Wiss. Gattingen, 4-5. Meijer, C. S., 1934: Nederl, Akad, Wetensch, Proc.,37, 85-812. Meijer, C. S., 1935a: Quart, J. Math., Oxford Ser. 6, 241-248, Meijer, C, S., 1935b: Nederl. Akad. Wetensch, Proc.,38, 528-535. Meijer, C. S., 1936: Nederd. Akad, Wetensch, Proc.,39, 394-403 and 519-527. Mefjer, C. S., 1936b: Math, Ann. 112, 469-499. Meijer, C. S., 1937: Neder, Akad. Wetensch, Proc.,40, 133-141 and 871-879, Meijer, C. S., 1938: Nederl, Akad, Wetensch,Proc., 41, 42-44, 275-277, 624-633, 744-755, 879-888 and 1108-1114. Meijer, C. S., 1941: Nederl. Akad, Wetensch, Proc.,44, 81-92, 186-194, 298-307, 436-451 and 590-598, Meixner, Joseph, 1933: Math, Z. 36, 677-707. Milne, Archibald, 1914/15: Proc. Edinburgh Math. Soc. 33, 46-64. Mitra, S. C., 1928: Tohoku Mash. j, 29, 321-325.CONFLUENT HYPERGEOMETRIC FUNCTIONS 295 REFERENCES Mitra, S. C., 1942: J. Indian Math, Soc, (N.S.) 6, 84-86 and 92-83. Mohan, Brij, 1943: Proc. Benares Math, Soc. 4, 59-60. Pasricha, B. R., 1943: Proc. Benares Math, Soc. 4, 61-69. Perron, Otto, 1921: J. Reine Angew. Math. (Credle) 151, 63-78, Pinney, Edmund, 1946: J. Math Physics 25, 49-79. Pochhammer, Leo, 1890: Math, Ann, 35, 495-526. Poole, E. G. C., 1935: Proc. London Math, Soc. (2) 38, 542-552. Poole, E. G. C., 1938: Quart, J. Math, Oxford Ser. 9, 230-233. Schmidt, Harry, 1937: J. Reine Angew. Math, (Crelle) 176, 250-252. Shanker, Hari, 1939: J. Indian Math, Soc. 3, 228+230. Shanker, Hari, 1943; Proc, Benares Math. Soc, 4, 51-57. Shanker, Hari, 1946: J. London Math, Soc. 21, 194-198. \ Sharma, Z. L. 1938: Philos. Mag, (7) 25, 491-504. errete Ser, J., 1937: Bull, Sei, Math, (2) 61, 74-81. Ser, J., 1938; Bull, Sci. Math. (2) 62, 171-182. Sexl, Theodor, 1929: Z. Physik 56, 62-93. Sommerfeld, Amold, 1939: Atombau und Spektrallinien 2° Aufl. tome 2, Friedrich Vieweg and Sohn, Brunswick, Szeg8, Gabor, 1939: Orthogonal Polynomials. Amer. Math. Soc. Colloquium Publications Vol, 23, New York. Taylor, W. C., 193% /. Math. Physics Mass, Inst, Tech. 18, 3449. Tricomi, Francesco, 1947: Ann. Mat, Pura Appl, (IV), 26, 141-175. Tricomi, Francesco, 1948: Equazioni differenziali, Turia, Ch. 5, section 7, 283- 300. Tricomi, Francesco, 1949: Anm Mat, Pura Apple (4) 28, 263-290. Tricomi, Francesco, 1950a: Math. Z, 52, 669-675. Tricomi, Francesco, 1950 b: Expansion of the hypergeometric function in series of cénfluent ones Tsvetkolf, G, E., 1941a: C. R. (Doklady) Acad. Sei, U.R.S.S. 32, 10-12. ‘Tsvetkolf, G. E., 1941b: C, Re Doklady) Acad. Sei, U.R.S.S. 33, 290-291. Watson, G, N., 1922: A treatise on the theory of Bessel functions, p. 100 MacMillan, Whittaker, E, T. and G. N. Watson, 1927: A course of modern analysis, Cambridge.SUBJECT INDEX All numbers refer to pages. Numbers in italics refer to the definitions. A ‘Automorphic functions, 98 B Basic hypergometric series, 195, 203 of several variables, 245 Bateman - Pasternack polynomials, 192 Bateman’s k-function, 267, 274 Bessel functions, as special case of G-function, 216 as special case of E-function, 204 connection with Meijer’s G-function, 216 connection with confluent hyper- geometric functions, 252, 265 expansion of confluent hyper geometric function in terms of 276 Bessel’s differential equation, 249 Beta function, 9 ff. (see also incomplete beta function) contour integrals for, 14 Bernoulli numbers,30, 35 expressed by zeta function, 34, 38 of higher arder, 39 Bemoulli polynomials, 36, 43 of higher order, 39 relation with generalized zeta function, 27 Binomial coefficients, relations involving, 85 Binomial series, generalization of, 196 truncated, 87, 101 Cc Confluence of singularities, 248 Confluent hypergeometric dit ential equation, 2488. fundamental systems of solutions, 258 general solation of, 253 transformation of, 251 Confluent hypergeometric functions, 248 associated, 254 asymptotic behavior of, 277 ff. behavior at origin, 262 branch cut for, 263 connection with Bessel functions, 265 contiguous, 254 derivative of, 254, 258 elementary relations for, 253, 257 expansions of, 275 integral addition theorem for, 271, 285 integrals involving, 283, 287 integral representations of, 255 ff, 2718. Laplace transforms containing, 269 logarithmic case, 260 multiplication formula for, 282 products of, 286 reduction of A, y" +A, y' + A,y=0, A =ax +, and other types of linear differential equation, to, 2498, relations between, 259 series involving, 284, 287 special values of parameters, 265 269 Wronskien, 253, 259 zeros of, 288 Conical functions, 174 Convergence, associated radii of, 227 296SUBJECT INDEX D Dougall-Ramanujan formula, 191 E E-function, (see MacRobert’s E -function) Elliptic functions, connection with basic hyper geometric series, 196 Elliptic modular functions, 99 Error functions, connection with confluent hypergeometric function, 266 Euler numbers, 40 of higher order, 43 Euler polynomials, 40, 44 of higher order, 43 Euler's constant, 1 Euler's dilogarithm, 31 Euler’s identity, 197 Euler's transformation of hyper- geometric function, 64, 253 Exponential integral, connection with confluent hypergeometric function, 266 Exponent differences, 90 F Fresnel imegrals, connection with confluent hyper- geometric function, 267 G 's G -function) Gamma function, 1 (see also incomplete gamma function) asymptotic expansions, 47 behavior of, at poles, 46 Binet’s expression for yy(z), 18, 21, 22 contour integrals for, 13 Dougall’s formula, 7 duplication formula for, 5 expansions for (14 2), (+2), T(z) and G(1 + 2), 45 expressions for log I'(z), 208. 48 functiona] equations for, 3 Gz), 20, 44 Gauss’ formula for ¥s(p/q), 18 297 infinite products expressed by, 5 infinite series involving, 7 infinite series in terms of yJ- function, 19 integrals containing, 49 Joncquiére’s relation for F (z, 5), 31 Kummer’s series, 23 Legendre’s duplication formula, 5 Lipschitz’s formula, 28 logarithmic derivative of, 15 ff., 23, 44 Multiplication formula of Gauss, 4 singularities of, 2 various notations for, 52 Gauss’ series (see hypergeometric series) Gegenbauer’s differential equation, 178 Gegenbauer’s functions, 178 Gegenbauer polynomials 175. Generalized hypergeometric differential equations, 184, 203, 210 Generalized hypergeometric functions (see generalized hypergeometric series) Generalized hypergeometric series, 182, 202 as special case of G-function, 215 basic, 195 contiguous, 187 convergence of, 182, 184 cubic transformations for , F,, 190 differential equations for, 184 Dixon’s theorem, 189 Dougall’s theorem, 189 {Py (@ ¢3 2), 248 ikedtities invovling, 185 integrals involving, 192 Vinear tansformations of || F , qm? 188, 190 nearly-poised, 188, 190 of unit argument, 183, 188, 190 Pochhammer -Barnes notation, 184 quadratic transformations for apr 190 recurrence relations for, 185 Saalschiitzian, 188 special cases of, 190, 192298 terminating, 182 Watson’s theorem, 189 Whipples’ theorem, 189 well-poised, 188, 190 H H -function, 204 Hankel-transform of , F, , 192 Heine’s hypergeometric series, 183 Bermite polynomials connection with confluent hyper geometric function, 267 Hypergeometric differential equation, 56 (see also hypergeometric series and functions; generalized hypergeometric series and func~ tions; hypergeometric functions of several variables) for Legendre polynomials, 122 solutions in the degenerate case of, 7 Hypergeometric equations, associated, 95 as special case of Riemann’s ‘equation, 91 degenerate case of, 69 degenerate solution of, 69 full solution, 74 group of, 93 singularities, 56 transformation of, 92 Hypergeometric function, 57 a relation for, F, (1,25 1+23— 1), 20 analytic continuation of, 62, 108 as continued fractions, 87 associated, 58 asymptotic expansions of, 75 bilinear relations for, 84 connection with Legendre functions, M1 contiguous, 57 cubic transformation of, 67, 113 degenerate case of, 68f. derivatives of, 58, 102 duplication formula for, 83 elementary functions as, 101 elementary relations for, 102 full solution of, 74 Gauss’ relations for contiguous, 103 SPECIAL FUNCTIONS generating function for F(b, a= b; ¢5 2), 82 Goursats’ tables of quadratic trans- formations for, 111 integrals involving, 79-81, 205 integral representations of, 59, 62, 67, 78, 114 Kummer’s relations between, 105 Legendre’s relation, 85 multiplication formulas for, 83 products of, 82 quadratic transformation of, 64, 110 121 special cases of, 89, 101, 122, 150 special values of, 104 transformations of higher degree of, 67 uniformization of, 99 zeros of, 99 Hypergeometric functions of several variables (see also hypergeometric series of several variables) analytic continuation of, 241 as solutions of partial differential equations, 231% Horn’s list, 224 integral representations of, 2298, reducible, 238 reduction of, to hypergeometric functions of one variable, 237 transformations of, 2398, Hypergeometric series, (see also hypergeometric functions) 87 convergence of, 57 terminating, 57 truncated, 192 Hypergeometric series of several variables, (see also hypergeometric function of several variables), 222 complete, 224 confluent, 225 convergence of, 227 expansion of, 243 Hom’s list of, 224 symbolic forms, 243 Hyperspherical harmonics, 245 1 Incomplete beta function, 87SUBJECT INDEX 299 Incomplete gamma functions, connection with confluent hyper- geometric function, 266 Integral cosine, connection with confluent hyper- geometric function, 266 Integral sine, connection with confluent hyper- geometric function, 266 7 Jacobi polynomiais, as special case of hypergeometric function, 81 bilinear generating function of, 244 generating function, 8] of two variables, 244 Jacobian theta functions, 99 K k-funetion, (see Bateman’s k function) Knar’s formula, 6 Kummer’s formulas for hypergeometric series, 64 Kummer’s series, for confluent hypergeometric function, 248 for log M(z), 23 of hypergeometric function, 92, 105 Kummer’s transformation, of confluent hypergeometric function, 253 of Whittaker’s function, 264 L Laguerre functions, connection with confluent hyper- geometric function, 268 Laguerre polynomials, connection with confluent hyper- geometric function, 268 expansion of confluent hypergeo- metric function in terms of, 275 Laplace integrals, for confluent hypergeometric function, 255 Laplace transforms, 269 of generalized hypergeometric series, 192 Laplace’s equation, 120, 173 Legendre’s differential equation, 12] as special case of hypergeometric equation, 122 Legendye’s functions, 122 addition theorem for, 168 associated, 148 asymptotic expansions for, 162 behavior of, at singular point, 163 derivatives of, 148, 181 Dougall’s expansion, 167 expansions in terms of, 165 Ferrer’s associated, 179 integral representations for, 1554. integrals involving, 169. Olbricht’s solutions of, 123 of special order and degree, 150 of integral order, 148 on the cut (for —1 hypergeometric series of two variables, 226 (s) Zeta function, 32 S,v), 24 &,s,v), 27 @(c,c;x) Confluent hypergeometric function, 248 @, 9B 5B (oer y) Confluent hypergeometric series of two variables, 225 , Basic hypergeometric series, * 195 w(z) Logarithmic derivative of gamma function, 15 Wa, c;x) Confluent hypergeometric function, 255 W,,¥, (01%) Confluent hyper- geometric series of two variables, 225 E+E, (+127) Confluent hyper- geometric series of two variables, 226 MISCELLANEOUS NOTATIONS arg z argument (or phase) of z (complex) Imz imaginary part of z (complex) Rez real part of z (complex) y Euler-Mascheroni constant (see p. 1) @) =D ata) (@) Tl (~aq”) @), 7 lal, 7 £ Cauchy principal value of an integral